Deep Neutral Networks (DNNs) were initially proposed towards the midst of the 20th century, motivated by the neural structure and mechanism of the human brain. Thanks to major ad-vancements in computational resources, during the past decade DNN-based systems have demonstrated unprecedented performance in terms of accuracy and speed. However, recent work has shown that such models may not be sufficiently robust during the inference process. Furthermore, due to the data-driven learning nature of DNNs, designing interpretable and gen-eralizable networks is a major challenge, especially in critical applications, such as medical Computer Aided Diagnostics (CAD) and other medical imaging tasks, including classification, regression and reconstruction. Within this context, a line of physics-driven approaches for deep learning has recently emerged, aimed at improving the stability and generalization capacity of DNNs for medical imaging applications. In this paper, we review recent work focused on phys-ics-driven or prior-information learning for a variety of imaging modalities and medical applica-tions. We discuss how the inclusion of domain-related knowledge into the learning process and networks’ design supports their stability and generalization capability. In addition, we highlight current and future challenges within this scope.

The transportation industry is rapidly transitioning from Internal Combustion Engine (ICE) based vehicles to Electric Vehicles (EVs) to promote clean energy. However, large-scale adoption of EVs can compromise the reliability of the power grids by introducing large uncertainty in the demand. Demand response with a controlled charge scheduling strategy for EVs can mitigate such issues. In this paper, a deep reinforcement learning- based charge scheduling strategy is developed for individual EVs by considering user’s dynamic driving behavior and charging preferences. The temporal dynamics of user’s anxiety about charging the EV battery is rigorously addressed. A dynamic weight allocation technique is applied to continuously tune user’s priority for charging and cost-saving with respect to charging duration. The sequential charging control problem is formulated as a Markov decision process, and an episodic approach to the deep deterministic policy gradient (DDPG) algorithm with target policy smoothing and delayed policy update techniques is applied to develop the optimal charge scheduling strategy. A real-world dataset that captures user’s driving behavior, such as arrival time, departure time, and charging duration, is utilized in this study. The extensive simulation results reveal the effectiveness of the proposed algorithm in minimizing energy cost while satisfying user’s charging requirements.

This last decade, the amount of data exchanged in the Internet increased by over a staggering factor of 100, and is expected to exceed well over the 500 exabytes by 2020. This phenomenon is mainly due to the evolution of high speed broadband Internet and, more specifically, the popularization and wide spread use of smartphones and associated accessible data plans. Although 4G with its long-term evolution (LTE) technology is seen as a mature technology, there is continual improvement to its radio technology and architecture such as in the scope of the LTE Advanced standard, a major enhancement of LTE. But for the long run, the next generation of telecommunication (5G) is considered and is gaining considerable momentum from both industry and researchers. In addition, with the deployment of the Internet of Things (IoT) applications, smart cities, vehicular networks, e-health systems, and Industry 4.0, a new plethora of 5G services has emerged with very diverging and technologically challenging design requirements. These include: high mobile data volume per area, high number of devices connected per area, high data rates, longer battery life for low-power devices, and reduced end-to-end latency. Several technologies are being developed to meet these new requirements. Among these we list ultra-densification, millimeter Wave usage, antennas with massive multiple-input multiple-output (MIMO), antenna beamforming to increase spacial diversity, edge/fog computing, among others. Each of these technologies brings its own design issues and challenges. For instance, ultra-densification and MIMO will increase the complexity to estimate channel condition and traditional channel state information (CSI) estimation techniques are no longer suitable due to the complexity of the new scenarios. As a result, new approaches to evaluate network condition such as by continuously collecting and monitoring key performance indicators become necessary. Timely decisions are needed to ensure the correct operation of such network. In this context, deep learning (DL) models could be seen as one of the main tools that can be used to process monitoring data and automate decisions. As these models are able to extract relevant features from raw data (images, texts, and other types of unstructured data), the integration between 5G and DL looks promising and one that requires exploring. As main contributions, this paper presents a systematic review about how DL is being applied to solve some 5G issues. We examine data from the last decade and the works that addressed diverse 5G problems, such as physical medium state estimation, network traffic prediction, user device location prediction, self network management, among others. We also discuss the main research challenges when using DL models in 5G scenarios and identify several issues that deserve further consideration.

Clinically, knowing the number of red blood cells (RBCs) and white blood cells (WBCs) helps doctors to make the better decision on accurate diagnosis of numerous diseases. The manual cell counting is a very time-consuming and expensive process, and it depends on the experience of specialists. Therefore, a completely automatic method supporting cell counting is a viable solution for clinical laboratories. This paper proposes a novel blood cell counting procedure to address this challenge. The proposed method adopts SegNet - a deep learning semantic segmentation to simultaneously segment RBCs and WBCs. The global accuracy of the segmentation of WBCs, RBCs, and the background of peripheral blood smear images obtains 89% when segment WBCs and RBCs from the background of blood smear images. Moreover, an effective solution to separate grouped or overlapping cells and cell count is presented using Euclidean distance transform, local maxima, and connected component labeling. The counting result of the proposed procedure achieves an accuracy of 93.3% for red blood cell count using dataset 1 and 97.38% for white blood cell count using dataset 2.

Recently, generative adversarial networks (GANs) can be used to generate the photo-realistic image from a low-dimension random noise. It is very dangerous that the synthesized or generated image is used on inappropriate contents in social media network. In order to successfully detect such fake image, an effective and efficient image forgery detector is desired. However, conventional image forgery detectors are failed to recognize the synthesized or generated images by using GAN-based generator since they are all generated but manipulation from the source. Therefore, we propose a deep learning-based approach to detect the fake image by combining the contrastive loss. First, several state-of-the-art GANs will be collected to generate the fake-real image pairs. Then, the contrastive will be used on the proposed common fake feature network (CFFN)to learn the discriminative feature between the fake image and real image (i.e., paired information). Finally, a smaller network will be concatenated to the CFFN to determine whether the feature of the input image is fake or real. Experimental results demonstrated that the proposed method significantly outperforms other state-of-the-art fake image detectors.

Advances in computing technology provide an opportunity to grow and develop an effective crop protection system. This study is a new way of diagnosing plant diseases using a deep and convolutional neural network. Deep learning is the strengthened method of machine learning. It uses a neural network that imitates like a human brain and gains certain information. Although powerful, its training process necessitates millions of tagged data points. The study's Methodology consists of three key stages: data acquisition, pre-processing, and Model Building.

The increasing use of deep learning techniques to manipulate images and videos, commonly referred to as "deepfakes," is making more and more challenging to differentiate between real and fake content. While various deepfake detection systems have been developed, they often struggle to detect deepfakes in real-world situations. In particular, these methods are often unable to effectively distinguish images or videos when these are modified using novel techniques which have not been used in the training set. In this study, we carry out an analysis of different deep learning architectures in an attempt to understand which is more capable of better generalizing the concept of deepfake. According to our results, it appears that Convolutional Neural Networks (CNNs) seem to be more capable of storing specific anomalies and thus excel in cases of datasets with a limited number of elements and manipulation methodologies. The Vision Transformer, conversely, is more effective when trained with more varied datasets, achieving more outstanding generalization capabilities than the other methods analysed. Finally, the Swin Transformer appears to be a good alternative for using an attention-based method in a more limited data regime. All the analyzed architectures seem to have a different way to look at deepfakes but since in a real-world environment, the generalization capability is essential, based on the carried out experiments the Vision Transformer seems to provide superior performances.

While visual assessment is the standard technique for burn evaluation, computer-aided diagnosis is increasingly sought due to high number of incidences globally. Patients are increasingly facing challenges which are not limited to shortage of experienced clinicians, lack of accessibility to healthcare facilities, and high diagnostic cost. Certain number of studies were proposed in discriminating burn and healthy skin using machine learning leaving a huge and important gap unaddressed; whether burns and related skin injuries can be effectively discriminated using machine learning techniques. Therefore, we specifically use pre-trained deep learning models due to deficient dataset to train a new model from scratch. Experiments were extensively conducted using three state-of-the-art pre-trained deep learning models that includes ResNet50, ResNet101 and ResNet152 for image patterns extraction via two transfer learning strategies: fine-tuning approach where dense and classification layers were modified and trained with features extracted by base layers, and in the second approach support vector machine (SVM) was used to replace top-layers of the pre-trained models, trained using off-the-shelf features from the base layers. Our proposed approach records near perfect classification accuracy of approximately 99.9%.

In the field of low power wireless networks, one of the techniques on which many researchers are putting their efforts is related to positioning methodologies such as fingerprinting in dense urban areas. This paper presents an experimental study aimed at quantifying the mean location estimation error in densely urbanized areas.Using a dataset made available by the University of Antwerp, a neural network was implemented with the aim of providing the position of the end-devices. In this way it was possible to measure the mean location estimation error in an area with high urban density. The results obtained show an accuracy in the localization of the end-device of less than 150 meters.This result would make it possible to use the fingerprint instead of alternative, energy consuming, methodologies such as GPS in IoT (Internet of Things) applications where battery life is the primary requirement to be met.

Fruit quality is a critical factor in the produce industry, affecting producers, distributors, consumers, and the economy. High-quality fruits are more appealing, nutritious, and safe, boosting consumer satisfaction and revenue for producers. Artificial Intelligence can aid in assessing the quality of the fruit using images. This paper presents a general machine-learning model for assessing fruit quality using deep image features. The model leverages the learning capabilities of the recent successful networks for image classification called Vision Transformers (ViT). The ViT model is built and trained with a combination of various fruit datasets and learned to distinguish between good and rotten fruit images. The general model demonstrated impressive results in accurately identifying the quality of various fruits such as Apples (with a 99.50% accuracy), Cucumbers (99%), Grapes (100%), Kakis (99.50%), Oranges (99.50%), Papayas (98%), Peaches (98%), Tomatoes (99.50%), and Watermelons (98%). However, it showed slightly lower performance in identifying Guavas (97%), Lemons (97%), Limes (97.50%), mangoes (97.50%), Pears (97%), and Pomegranates (97%).

The emergence of unmanned aerial vehicles (UAVs) raised multiple concerns, given their potential for malicious misuse in unlawful acts Vision-based counter-UAV applications offer a reliable solution compared to acoustic and radio frequency-based solutions because of their high detection accuracy in diverse weather conditions. The existing solutions work well on trained datasets, but their accuracy is relatively low for real-time detection. In this paper, we model deep learning-empowered solutions to improve the multi-class UAV's classification performance using single-shot object detection algorithms YOLOv5 and YOLOv7. The transfer learning is employed for performance improvement and rapid training with improved results. We customized a multi-class dataset containing multi-rotor, fixed-wing, and single-rotor UAVs in challenging weather conditions. Experiments show that the integration of transfer learning has achieved good results, with an overall best average-classification precision of 94\%, an average recall of 93.1\%, a mAP$@$0.5 average of 95.3\%, and an average F1 score of 92.33\%. The dataset and code are available as an open source: https://github.com/ZeeshanKaleem/YOLOV5-Large-vs-YOLOV7.git

Introduction: Dental segmentation in panoramic radiograph has become very relevant in dentistry, since it allows health professionals to carry out their assessments more clearly and helps them to define the best possible treatment plan for their patients. Objectives: In this work, a comparative study is carried out with four segmentation algorithms (U-Net, DCU-Net, DoubleU-Net and Nano-Net) that are prominent in the medical literature on segmentation and we evaluate their results with the current state of the art of dental segmentation in panoramic radiograph. Methods: These algorithms were tested with a dataset consisting of 1,500 images, considering experiment scenarios with and without augmentation data. Results: DoubleU-Net was the model that presented the best results among the analyzed models, reaching 96.591% accuracy and 92.886% Dice using the dataset with data augmentation. Another model that stood out was Nano-Net using the dataset without data augmentation; this model achieved results close to that of the literature with only 235 thousand trainable parameters, while the literature model (TSASNet) contains 78 million. Conclusions: The results obtained in this work are satisfactory and present paths for a better and more effective dental segmentation process.

Beyond semantic segmentation,3D instance segmentation(a process to delineate objects of interest and also classifying the objects into a set of categories) is gaining more and more interest among researchers since numerous computer vision applications need accurate segmentation processes(autonomous driving, indoor navigation, and even virtual or augmented reality systems…) This paper gives an overview and a technical comparison of the existing deep learning architectures in handling unstructured Euclidean data for the rapidly developing 3D instance segmentation. First, the authors divide the 3D point clouds based instance segmentation techniques into two major categories which are proposal based methods and proposal free methods. Then, they also introduce and compare the most used datasets with regard to 3D instance segmentation. Furthermore, they compare and analyze these techniques performance (speed, accuracy, response to noise…). Finally, this paper provides a review of the possible future directions of deep learning for 3D sensor-based information and provides insight into the most promising areas for prospective research.

Today, specific CNN models assigned to specific tasks are often used. In this article, the authors have investigated three models MobileNet, EfficientNetB0 and InceptionV3. Artificial intelligence methods for deep networks will be presented but implemented in a limited computer resource. Three types of training bases were investigated, starting with a simple base verifying five colors, followed by recognition of two different orthogonal elements, and then more complex images discriminating between elements. The research aimed to demonstrate the models' capabilities based on training base parameters such as the number of images and epoch types. The architectures proposed by the authors in these cases were chosen based on simulation studies conducted on a virtual machine with limited hardware parameters. The proposals present the advantages and disadvantages of different models based on TensorFlow, Keras libraries in the Jupiter environment written based on the Python language. The authors' proposal of the selected AI model enables further work on e.g. image classification, but in a limited computer resource for industrial implementation e.g. on a PLC.

Recently, the use of drones/unmanned aerial vehicles (UAVs) has notably increased due to their broad commercial spread and low cost. The wide diffusion of drones increases the hazards of their misuse in illegitimate actions such as drug smuggling and terrorism. Thus, the surveillance and automated detection of drones are crucial for safeguarding restricted regions or special zones from illegal drone interventions. One of the most challenging issues in drone detection in surveillance videos is the apparent similarity of drones and birds against complex backgrounds. In this work, an automated image-based drone-detection system utilizing an advanced deep-learning-based object-detection method known as you only look once (YOLOv5) is introduced for protecting restricted regions or special zones from unlawful drone interventions. Due to the lack of sufficient data, transfer learning was utilized to pretrain the object-detection method to increase the performance. The experiments showed outstanding results, and an average precision of 94.7% was accomplished.

Scientists and astronomers have attached Scientists and astronomers have attached great importance to the task of discovering new exoplanets, even more so if they are in the habitable zone. To date, more than 4300 exoplanets have been confirmed by NASA, using various discovery techniques, including planetary transits, in addition to the use of various databases provided by space and ground-based telescopes. This article proposes the development of a deep learning system for detecting planetary transits in Kepler Telescope lightcurves. The approach is based on related work from the literature and enhanced to validation with real lightcurves. A CNN classification model is trained from a mixture of real and synthetic data, and validated only with real data and different from those used in the training stage. The best ratio of synthetic data is determined by the perform of an optimisation technique and a sensitivity analysis. The precision, accuracy and true positive rate of the best model obtained are determined and compared with other similar works. The results demonstrate that the use of synthetic data on the training stage can improve the transit detection performance on real light curves.

The explosion of data has transformed the world since much more information is available for collection and analysis than ever before. To extract valuable information from the data in different dimensions, various deep learning models have been developed in the past years. Although these models have demonstrated their strong capability on improving products and services in various applications, training them is still a time-consuming and resource-intensive process. Presently, cloud, one of the most powerful computing infrastructures, has been used for the training. However, how to manage cloud computing resources and to perform the training efficiently is still challenging current techniques. For example, general resource scheduling approaches, such as spread priority and balanced resource schedulers, actually do not work well with deep learning workloads. Besides, the resource allocation problem on a cluster can be divide into two subproblems: (1) local resource optimization: improve resource configuration for a single machine; (2) global resource optimization: improve the cluster-wide resource allocation. In this thesis, we propose two novel container schedulers, FlowCon and SpeCon, that are designed to address these two subproblems respectively and specifically to optimize performance of short-lived deep learning applications in the cloud. FlowCon focuses on resource configuration of single-node in a cluster, as show that it efficiently improves deep learning tasks completion time and resource utilization, and reduces the completion time of a specific job by up to 42.06\% without sacrificing the overall system time. SpeCon targets on cluster-wide resource configuration that speculatively migrate slow-growing models to release resources for fast-growing ones. Based on our experiments, SpeCon improves makespan for up to 24.7\%, compared to current approaches.

This paper presents a two-dimensional mathematical model of compound eye vision. Such a model is useful for solving navigation issues for autonomous mobile robots on the ground plane. The model is inspired by the insect compound eye that consists of ommatidia, which are tiny independent photoreception units, each of which combines a cornea, lens, and rhabdom. The model describes the planar binocular compound eye vision, focusing on measuring distance and azimuth to a circular feature with an arbitrary size. The model provides a necessary and sufficient condition for the visibility of a circular feature by each ommatidium. On this basis, an algorithm is built for generating a training data set to create two deep neural networks (DNN): the first detects the distance, and the second detects the azimuth to a circular feature. The hyperparameter tuning and the configurations of both networks are described. Experimental results showed that the proposed method could effectively and accurately detect the distance and azimuth to objects.

Oil spillage over a sea or ocean’s surface is a threat to marine and coastal ecosystems. Spaceborne synthetic aperture radar (SAR) data has been used efficiently for the detection of oil spills due to its operational capability in all-day all-weather conditions. The problem is often modeled as a semantic segmentation task. The images need to be segmented into multiple regions of interest such as sea surface, oil spill, look-alikes, ships and land. Training of a classifier for this task is particularly challenging since there is an inherent class imbalance. In this work, we train a convolutional neural network (CNN) with multiple feature extractors for pixel-wise classification; and introduce to use a new loss function, namely ‘gradient profile’ (GP) loss, which is in fact the constituent of the more generic Spatial Profile loss proposed for image translation problems. For the purpose of training, testing and performance evaluation, we use a publicly available dataset with selected oil spill events verified by the European Maritime Safety Agency (EMSA). The results obtained show that the proposed CNN trained with a combination of GP, Jaccard and focal loss functions can detect oil spills with an intersection over union (IoU) value of 63.95%. The IoU value for sea surface, look-alikes, ships and land class is 96.00%, 60.87%, 74.61% and 96.80%, respectively. The mean intersection over union (mIoU) value for all the classes is 78.45%, which accounts for a 13% improvement over the state of the art for this dataset. Moreover, we provide extensive ablation on different Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) based hybrid models to demonstrate the effectiveness of adding GP loss as an additional loss function for training. Results show that GP loss significantly improves the mIoU and F1 scores for CNNs as well as ViTs based hybrid models. GP loss turns out to be a promising loss function in the context of deep learning with SAR images.

Prediction of stock groups values has always been attractive and challenging for shareholders. This paper concentrates on the future prediction of stock market groups. Four groups named diversified financials, petroleum, non-metallic minerals and basic metals from Tehran stock exchange are chosen for experimental evaluations. Data are collected for the groups based on ten years of historical records. The values predictions are created for 1, 2, 5, 10, 15, 20 and 30 days in advance. The machine learning algorithms utilized for prediction of future values of stock market groups. We employed Decision Tree, Bagging, Random Forest, Adaptive Boosting (Adaboost), Gradient Boosting and eXtreme Gradient Boosting (XGBoost), and Artificial neural network (ANN), Recurrent Neural Network (RNN) and Long short-term memory (LSTM). Ten technical indicators are selected as the inputs into each of the prediction models. Finally, the result of predictions is presented for each technique based on three metrics. Among all algorithms used in this paper, LSTM shows more accurate results with the highest model fitting ability. Also, for tree_based models, there is often an intense competition between Adaboost, Gradient Boosting and XGBoost.

Tropical cyclones (TC) are dangerous weather events and accurate monitoring and forecasting can provide significant early warning to reduce loss of life and property. However, the study of tropical cyclone intensity remains challenging, both in terms of theory and forecasting. ERA5 reanalysis is benchmark data set for tropical cyclone studies, yet the maximum wind speed error is very large (68 kts) and still 19 kts after simple linear correction even in the better sampled North Atlantic. Here, we develop an adaptive learning approach to correct the intensity in the ERA5 reanalysis, by optimising the inputs to overcome the problems because of the poor data quality and updating the features to improve the generalisability of the deep learning-based model. Specifically, we use TC knowledge to increase the representativeness of the inputs so that the general features can be learned with deep neural networks in the sample space, and then use domain adaptation to update the general features from the known domain with historical storms to the specific features for the unknown domain of new storms. This approach can reduce the error to only 6 kts which is within the uncertainty of the best track data in IBTrACS in the North Atlantic. The method may have wide applicability, such as extending it to the correction of intensity estimation from satellite imagery and intensity prediction from dynamical models.

Sparse reward long horizon task is a major challenge for deep reinforcement learning algorithm. One of the key barriers is data-inefficiency. Even in the simulation environment, it usually takes weeks to training the agent. In this study, a data-efficiency training framework is proposed, where a curriculum learning is design for the agent in the simulation scenario. Different distributions of the initial state are set for the agent to get more informative reward during the whole training process. A fine-tuning of the parameters in the output layer of the neural network for value function is conduct to bridge the gap between sim-to-real. An experiment of UAV maneuver control is conducted in the proposed training framework to verify the method more efficient. We demonstrate that data-efficiency is different for the same data in different training stages.

In this paper, we propose a new methodology for crack monitoring in concrete structures. This approach is based on a n this paper, we propose a new methodology for monitoring cracks in concrete structures. This approach is based on a multi-resolution analysis of a sample or a specimen of the studied material subjected to several types of solicitation. The image obtained by ultrasonic investigation and processing by a dedicated wavelet will be analyzed according to several scales in order to detect internal cracks and crack initiation. The ultimate goal of this work is to propose an automatic crack type identification scheme based on convolutional neural networks (CNN). In this context, crack propagation can be monitored without access to the concrete surface and the goal is to detect cracks before they are visible on the concrete surface. The key idea allowing such a performance is the combination of two major data analysis tools which are wavelets and Deep Learning. This original procedure allows to reach a high accuracy close to 0.90. In this work, we have also implemented another approach for automatic detection of external cracks by deep learning from publicly available datasets.

Deep learning (DL) algorithms have recently emerged from machine learning and soft computing techniques. Since then, several deep learning (DL) algorithms have been recently introduced to scientific communities and are applied in various application domains. Today the usage of DL has become essential due to their intelligence, efficient learning, accuracy and robustness in model building. However, in the scientific literature, a comprehensive list of DL algorithms has not been introduced yet. This paper provides a list of the most popular DL algorithms, along with their applications domains.

Melanoma is widely recognized as one of the most lethal forms of skin cancer, with its incidence showing an upward trend in recent years. Nonetheless, the timely detection of this malignancy substantially enhances the likelihood of patients’ long-term survival. Several computer-based methods have recently been proposed in the pursuit of diagnosing skin lesions at their early stages. Despite achieving some level of success, there still remains a margin of error that the machine learning community considers to be an unresolved research challenge. This study presents a novel framework for the classification of skin lesions. The framework incorporates deep features to generate a highly discriminant feature vector, while also maintaining the integrity of the original feature space. Recent deep models including Darknet53, DenseNet201, InceptionV3, and InceptionResNetV2 are employed in our study for the purpose of feature extraction. Additionally, transfer learning is leveraged to enhance the performance of our approach. In the subsequent phase, the extracted feature information from the chosen pre-existing models is combined, with the aim of preserving maximum information, prior to undergoing the process of feature selection using a novel entropy-controlled grey wolf optimization (ECGWO) algorithm. The integration of fusion and selection techniques is employed to initially incorporate the feature vector with a high level of information and subsequently eliminate redundant and irrelevant feature information. The efficacy of our design is substantiated through the evaluation on three benchmark dermoscopic datasets, namely PH2, ISIC-MSK, and ISIC-UDA. In order to validate the proposed methodology, a comprehensive evaluation is conducted, including a rigorous comparison with established techniques in the field.

This paper develops a multi-task learning framework that attempts to incorporate the image structure knowledge to assist image inpainting, which is not well explored in previous works. The primary idea is to train a shared generator to simultaneously complete the corrupted image and corresponding structures --- edge and gradient, thus implicitly encouraging the generator to exploit relevant structure knowledge while inpainting. In the meantime, we also introduce a structure embedding scheme to explicitly embed the learned structure features into the inpainting process, thus to provide possible preconditions for image completion. Specifically, a novel pyramid structure loss is proposed to supervise structure learning and embedding. Moreover, an attention mechanism is developed to further exploit the recurrent structures and patterns in the image to refine the generated structures and contents. Through multi-task learning, structure embedding besides with attention, our framework takes advantage of the structure knowledge and outperforms several state-of-the-art methods on benchmark datasets quantitatively and qualitatively.

Deep Learning is an extremely important research topic in Earth Observation. Current use-cases range from semantic image segmentation, object detection to more common problems found in computer vision such as object identification. Earth Observation is an excellent source for different types of problems and data for Machine Learning in general and Deep Learning in particular. It can be argued that both Earth Observation and Deep Learning as fields of research will benefit greatly from this recent trend of research. In this paper we take several state of the art Deep Learning network topologies and provide a detailed analysis of their performance for semantic image segmentation for building footprint detection. The dataset used is comprised of high resolution images depicting urban scenes. We focused on single model performance on simple RGB images. In most situations several methods have been applied to increase the accuracy of prediction when using deep learning such as ensembling, alternating between optimisers during training and using pretrained weights to bootstrap new models. These methods although effective, are not indicative of single model performance. Instead, in this paper, we present different topology variations of these state of the art topologies and study how these variations effect both training convergence and out of sample, single model, performance.

If the development of machine learning and artificial intelligence plays a role in many fields of research and technology today, it has a special relationship with neurosciences. Indeed, historically inspired by our knowledge of the brain, deep learning shares some vocabularies with neurosciences and can sometimes be considered a brain’s model. Taking the particular example of epileptic seizure, which can develop in any biological neural tissue, we raise the question if and how the models used for deep learning can capture or model these pathological events. This particular example is a starting point to discuss the nature, limits, and functions of these models, and what we expect from a model of the brain. Finally, we argue that a pluralistic approach leading to the integrated coexistence of different models is necessary to study the brain in all its complexity.

Rainfall retrieval using geostationary satellites provides critical means to the monitoring of extreme rainfall events. Using the relatively new Himawari 8 meteorological satellite with three times more channels than its predecessors, the deep learning framework of “convolutional autoencoder” (CAE) was applied to the extraction of cloud and precipitation features. The CAE method was incorporated into the Convolution Neural Network version of the PERSIANN precipitation retrieval that uses GOES satellites. By applying the CAE technique with the addition of Residual Blocks and other modifications of deep learning architecture, the presented derivation of PERSIANN operated at the Central Weather Bureau of Taiwan (referred to as PERSIANN-CWB) expands four extra convolution layers to fully use Himawari 8’s infrared and water vapor channels, while preventing degradation of accuracy caused by the deeper network. The development of PERSIANN-CWB was trained over Taiwan for its diverse weather systems and localized rainfall features, and the evaluation reveals an overall improvement from its CNN counterpart and superior performance over all other rainfall retrievals analyzed. Limitation of this model was found in the derivation of typhoon rainfall, an area requiring further research.

The increasing trend of studying the innate softness of robotic structures and amalgamating it with the benefits of the extensive developments in the field of embodied intelligence has led to sprouting of a relatively new yet extremely rewarding sphere of technology. The fusion of current deep reinforcement algorithms with physical advantages of a soft bio-inspired structure certainly directs us to a fruitful prospect of designing completely self-sufficient agents that are capable of learning from observations collected from their environment to achieve a task they have been assigned. For soft robotics structure possessing countless degrees of freedom, it is often not easy (something not even possible) to formulate mathematical constraints necessary for training a deep reinforcement learning (DRL) agent for the task in hand, hence, we resolve to imitation learning techniques due to ease of manually performing such tasks like manipulation that could be comfortably mimicked by our agent. Deploying current imitation learning algorithms on soft robotic systems have been observed to provide satisfactory results but there are still challenges in doing so. This review article thus posits an overview of various such algorithms along with instances of them being applied to real world scenarios and yielding state-of-the-art results followed by brief descriptions on various pristine branches of DRL research that may be centers of future research in this field of interest.

We examine the use of deep learning (neural networks) to predict the movement of the S&P 500 Index using past returns of all the stocks in the index. Our analysis finds that the future direction of the S&P 500 index can be weakly predicted by the prior movements of the underlying stocks in the index. Decomposition of the prediction error indicates that most of the lack of predictability comes from randomness and only a little from nonstationarity. We believe this is the first test of S&P500 market efficiency that uses a very large information set, and it extends the domain of weak-form market efficiency tests.

Chinese calligraphy, revered globally for its therapeutic and mindfulness benefits, encompasses styles such as Regular (Kai Shu), Running (Xing Shu), Official (Li Shu), and Cursive (Cao Shu) scripts. Beginners often start with Regular script, advancing to more intricate styles like Cursive. Each style, marked by unique historical calligraphers' contributions, requires learners to discern distinct nuances. The integration of AI in calligraphy analysis, collection, recognition, and classification are pivotal. This study introduces an innovative Convolutional Neural Network (CNN) architecture, pioneering the application of CNN in the classification of Chinese calligraphy. Focusing on the four principal calligraphers' styles from the Tang dynasty (690-907 A.D), this research spotlights the era when the traditional regular script font (Kai Shu) was refined. A comprehensive dataset of 8282 samples from these calligraphers, representing the zenith of regular style, was compiled for CNN training and testing. The model distinguishes personal styles for classification, showing superior performance over existing networks. Achieving 89.5-96.2% accuracy in calligraphy classification, our approach underscores the significance of CNN in both font and artistic style categorization. This research paves the way for advanced studies in Chinese calligraphy and its cultural implications.

Alzheimer's is a disease for which there is no cure. Diagnosing Alzheimer's Disease (AD) early facilitates family planning and cost control. The purpose of this study is to predict the presence of AD using socio-demographic, clinical, and Magnetic Resonance Imaging (MRI) data. Early detection of AD enables family planning and may reduce costs by delaying long-term care. Accurate, non-imagery methods also reduce patient costs. The Open Access Series of Imaging Studies (OASIS-1) cross-sectional MRI data were analyzed. A gradient boosted machine (GBM) predicted the presence of AD as a function of gender, age, education, socioeconomic status (SES), and Mini-Mental State Exam (MMSE). A Residual Network with 50 layers (ResNet-50) predicted CDR presence and severity from MRI's (multi-class classification). The GBM achieved a mean 91.3% prediction accuracy (10-fold stratified cross validation) for dichotomous CDR using socio-demographic and MMSE variables. MMSE was the most important feature. ResNet-50 using image generation techniques based on an 80% training set resulted in 98.99% three class prediction accuracy on 4,139 images (20% validation set) at Epoch 133 and nearly perfect multi-class predication accuracy on the training set (99.34%). Machine Learning methods classify AD with high accuracy. GBM models may help provide initial detection based on non-imagery analysis, while ResNet-50 network models might help identify AD patients automatically prior to provider review.

Retailers must have accurate sales forecasts to operate their businesses efficiently and effectively and to remain competitive in the marketplace. Global forecasting models like RNNs can be a powerful tool for forecasting in retail settings, where multiple time series are often interrelated and influenced by a variety of external factors. By including covariates in a forecasting model, we can better capture the various factors that can influence sales in a retail setting. This can help improve the accuracy of our forecasts and enable better decision-making for inventory management, purchasing, and other operational decisions. In this study we investigate how the accuracy of global forecasting models is affected by the inclusion of different potential demand covariates. To ensure the significance of the study's findings, we used the M5 forecasting competition's openly accessible and well-established dataset. The results obtained from the DeepAR models, which were trained on various combinations of features including time, price, events, and IDs, suggest that individually only the features corresponding to IDs improve the baseline model. However, when all the features are used together, the best performance is achieved, indicating that the individual relevance of each feature is emphasized when the information is given jointly. Comparing the model with features to the model without features, there is an improvement of 1.76\% for MRMSSE and 6.47\% for MMASE.

Sometimes when diabetic retinopathy (DR) is found and treated quickly, vision loss can indeed be spared. This study deploys a deep learning (DL) model that can discover all 5 stages of DR more accurately than other methods. The proposed methodology shows two cases scenarios: case 1 with image enhancement using CLAHE and ESRGAN, and case 2 without image enhancement. Augmentation techniques are then employed to produce a balanced dataset with the identical criteria for both scenarios. The generated model using DenseNet-121 on the APTOS dataset outperformed other approaches for locating the 5 stages of DR, with an accuracy of 98.7 percent for case 1 and 81.2 percent for case 2. Using CLAHE and ESRGAN was shown to improve a model's performance and ability to learn.

Online social networks (OSNs) play an integral role in facilitating social interaction; however, these social networks increase antisocial behavior, such as cyberbullying, hate speech, and trolling. Aggression or hate speech that takes place through short message service (SMS) or the Internet (e.g., in social media platforms) is known as cyberbullying. Therefore, automatic detection utilizing natural language processing (NLP) is a necessary first step that helps prevent cyberbullying. This research proposes an automatic cyberbullying method to detect aggressive behavior using a consolidated deep learning model. This technique utilizes multichannel deep learning based on three models, namely, the bidirectional gated recurrent unit (BiGRU), transformer block, and convolutional neural network (CNN), to classify Twitter comments into two categories: aggressive and not aggressive. Three well-known hate speech datasets were combined to evaluate the performance of the proposed method. The proposed method achieved promising results. The accuracy of the proposed method was approximately 88%.

This paper proposes a neural network which fuses the data received from a camera system on a gantry, to detect moving objects and calculate relative position and velocity of the vehicles traveling on a freeway, this information is used to estimate the traffic flow. To estimate the traffic flow at both microscopic and macroscopic view, this paper used YOLO v4 and DeepSORT for vehicle detection and tracking, then counting the number of vehicles pass through the freeway by drawing virtual lines and hot zones, also counting the velocity of each vehicles. The information is then pass to the traffic control center, in order to monitoring and control traffic flow on freeways, and analyzing freeway conditions.

Learning to maintain postural balance while standing requires a significant fine coordination effort between the neuromuscular system and the sensory system. It is one of the key contributing factors towards fall prevention, especially in the older population. Using artificial intelligence (AI), we can similarly teach an agent to maintain a standing posture, and thus teach the agent not to fall. In this paper, we investigate the learning progress of an AI agent and how it maintains a stable standing posture through reinforcement learning. During training, the AI agent learnt three policies. First, it learnt to maintain the Centre-of-Gravity and Zero-Moment-Point in front of the body. Then, it learnt to shift the load of the entire body on one leg while using the other leg for fine tuning the balancing action. Finally, it started to learn the coordination between the two pre-trained policies. This study shows the potentials of using deep reinforcement learning in human movement studies. The learnt AI behaviour also exhibited attempts to achieve an unplanned goal because it correlated with the set goal (e.g. walking in order to prevent falling). The failed attempts to maintain a standing posture is an interesting by-product which can enrich the fall detection and prevention research efforts.

Now when the whole world is still under COVID-19 pandemic, many schools have transferred the teaching from physical classroom to online platforms. It is highly important for schools and online learning platforms to investigate the feedback to get valuable insights about online teaching process so that both platforms and teachers are able to learn which aspect they can improve to achieve better teaching performance. But handling reviews expressed by students would be a pretty laborious work if they were handled manually as well as it is unrealistic to handle large-scale feedback from e-learning platform. In order to address this problem, both machine learning algorithms and deep learning models are used in recent research to automatically process students' review getting the opinion, sentiment and attitudes expressed by the students. Such studies may play a crucial role in improving various interactive online learning platforms by incorporating automatic analysis of feedback. Therefore, we conduct an overview study of sentiment analysis in educational field presented in recent research, to help people grasp an overall understanding of the sentiment analysis research. Besides, according to the literature review, we identify three future directions that researchers can focus on in automatically feedback processing: high-level entity extraction, multi-lingual sentiment analysis, and handling of figurative language.

In order for robots to attain more general-purpose utility, grasping is a necessary skill to master. Such general-purpose robots may use their perception abilities in order to visually identify grasps for a given object. A grasp describes how a robotic end-effector can be arranged on top of an object to securely grab it between the robotic gripper and successfully lift it without slippage. Traditionally, grasp detection requires expert human knowledge to analytically form the task-specific algorithm, but this is an arduous and time-consuming approach. During the last five years, deep learning methods have enabled significant advancements in robotic vision, natural language processing, and automated driving applications. The successful results of these methods have driven robotics researchers to explore the application of deep learning methods in task generalised robotic applications. This paper reviews the current state-of-the-art in regards to the application of deep learning methods to generalised robotic grasping and discusses how each element of the deep learning approach has improved the overall performance of robotic grasp detection. A number of the most promising approaches are evaluated and the most successful for grasp detection is identified as the one-shot detection method. The availability of suitable volumes of appropriate training data is identified as a major obstacle for effective utilisation of the deep learning approaches, and the use of transfer learning techniques is identified as a potential mechanism to address this. Finally, current trends in the field and future potential research directions are discussed.

This paper analyses the application of deep learning techniques for predicting wave overtopping events in port environments using sea state and weather forecasts as inputs. The study was conducted in the outer port of Punta Langosteira, A Coruña, Spain. A video recording infrastructure was installed to monitor overtopping events from 2015 to 2022, identifying 3709 overtopping events. The data collected was merged with actual and predicted data for the sea state and weather conditions during the overtopping events, creating three datasets. We used these datasets to create several machine learning models to predict whether an overtopping event would occur based on sea state and weather conditions. The final models achieved a high accuracy level during the training and testing stages: 0.81, 0.73, and 0.84 average accuracy during training and 0.67, 0.48, and 0.86 average accuracy during testing, respectively. The results of this study have significant implications for port safety and efficiency, as wave overtopping events can cause disruptions and potential damage. Using deep learning techniques for overtopping prediction can help port managers take preventative measures and optimize operations, ultimately improving safety and helping to minimize the economic impact that overtopping events have on the port's activities.

Machine learning is a fundamental aspect of artificial intelligence that involves the development of algorithms and models that allow computers to learn and make predictions or decisions without explicit programming. With the development of neural networks, back-propagation algorithms and deep learning, machine learning has made breakthroughs in the fields of image recognition, natural language processing and handwriting recognition using machine learning techniques. The advent of deep learning has revolutionized the field of handwriting recognition, using convolutional neural networks, recurrent neural networks, and sequence-to-sequence models to provide solutions that go beyond machine learning methods and significantly improve the accuracy and robustness of handwriting recognition systems. But challenges remain, including the need for large labelled datasets, computational resources and addressing potential biases. As research in deep learning techniques continues to drive handwriting recognition closer towards realisability, machine learning approaches remain at the forefront.

This is an article or technical note which is intended to provides an insight journey of Machine Learning Systems (MLS) testing, its evolution, current paradigm and future work. Machine Learning Models, used in critical applications such as healthcare industry[1], Automobile [2], [3] and Air Traffic control, Share Trading etc., and failure of ML Model can lead to severe consequences in terms of loss of life or property. To remediate this, developers, scientists, and ML community around the world, must build a highly reliable test architecture for critical ML application. At the very foundation layer, any test model must satisfy the core testing attributes such as test properties and its components. This attribute comes from the software engineering [5], [6], but the same cannot be applied in as-is form to the ML testing and we will tell you “why”.

A hybrid smart process and material design system for nanoimprinting is proposed, which combines with a learning system based on experimental and numerical simulation results. Instead of carrying out extensive learning experiments for various condition, the simulation learning results are partially complimented where the results can be theoretically predicted by numerical simulation. In other word, the lacking data in experimental learning are complimented by simulation-based learning results. Therefore, the prediction of nanoimprint results under various conditions without experimental learning could be realized even for unknown materials. In this study, material and process design for a low-temperature nanoimprint process using glycerol-containing polyvinyl alcohol are demonstrated. Experimental results under limited conditions are learned to investigate optimum Glycerol concentrations and process temperatures. On the other hand, simulation-based learning is used to predict the dependence on press pressure and shape parameters. The prediction results for unknown Glycerol concentrations agreed well with the follow-up experiments.

Olive fruits at different ripening stages give rise to various table olive products and oil qualities. Therefore, developing an efficient method for recognizing and sorting olive fruits based on their ripening stages can greatly facilitate postharvest processing. This study introduces an automatic computer vision system that utilizes deep learning technology to classify the `Roghani` Iranian olive cultivar into five ripening stages using color images. The developed model employs convolutional neural networks (CNN) and transfer learning based on the Xception architecture and ImageNet weights as the base network. The model was fine-tuned by testing multiple configurations of well-known CNN layers. To minimize overfitting and enhance model generality, data augmentation techniques were employed. By considering different optimizers and two image sizes, four final candidate models were generated. These models were then compared in terms of loss and accuracy on the test dataset, classification performance (classification report and confusion matrix), and generality. All four candidates exhibited high accuracies ranging from 86.93% to 93.46% and comparable classification performance. In all models, at least one class was recognized with 100% accuracy. However, by taking into account the risk of overfitting, two models were discarded. Finally, a model with an image size of 224 × 224 and an SGD optimizer, which had a loss of 1.23 and an accuracy of 86.93%, was selected as the preferred option. The results of this study offer robust tools for automatic olive sorting systems, simplifying the differentiation of olives at various ripening levels for different post-harvest products.

Early detection improves prognosis in pancreatic ductal adenocarcinoma (PDAC) but is challenging as lesions are often small and poorly defined on contrast-enhanced computed tomography scans (CE-CT). Deep learning can facilitate PDAC diagnosis, however current models still fail to identify small (<2cm) lesions. In this study, state-of-the-art deep learning models were used to develop an automatic framework for PDAC detection, focusing on small lesions. Additionally, the impact of integrating surrounding anatomy was investigated. CE-CT scans from a cohort of 119 pathology-proven PDAC patients and a cohort of 123 patients without PDAC were used to train a nnUnet for automatic lesion detection and segmentation (nnUnet_T). Two additional nnUnets were trained to investigate the impact of anatomy integration: (1) segmenting the pancreas and tumor (nnUnet_TP), (2) segmenting the pancreas, tumor, and multiple surrounding anatomical structures (nnUnet_MS). An external, publicly available test set was used to compare the performance of the three networks. The nnUnet_MS achieved the best performance, with an area under the receiver operating characteristic curve of 0.91 for the whole test set and 0.88 for tumors <2cm, showing that state-of-the-art deep learning can detect small PDAC and benefits from anatomy information.

In recent years, a range of problems under the broad umbrella of computer vision based analysis of ancient coins have been attracting an increasing amount of attention. Notwithstanding this research effort, the results achieved by the state of the art in published literature remain poor and far from sufficiently well performing for any practical purpose. In the present paper we present a series of contributions which we believe will benefit the interested community. We explain that the approach of visual matching of coins, universally adopted in existing published papers on the topic, is not of practical interest because the number of ancient coin types exceeds by far the number of those types which have been imaged, be it in digital form (e.g., online) or otherwise (traditional film, in print, etc.). Rather, we argue that the focus should be on understanding the semantic content of coins. Hence, we describe a novel approach—to first extract semantic concepts from real-world multimodal input and associate them with their corresponding coin images, and then to train a convolutional neural network to learn the appearance of these concepts. On a real-world data set, we demonstrate highly promising results, correctly identifying a range of visual elements on unseen coins with up to 84% accuracy.

We present a hierarchical reinforcement learning (RL) architecture that employs various low-level agents to act in the trading environment, i.e. the market. The highest level agent selects among a group of specialised agents, and then the selected agent decides when to sell or buy a single asset for some period. This period can be variable according to a termination function. We hypothesized that due to different market regimes, more than one single agent is needed when trying to learn from such heterogeneous data, and instead, multiple agents will perform better, with each one specialising in a subset of the data. We use $k-means$ clustering to partition the data and train each agent with a different cluster. Partitioning the input data also helps model-based RL (MBRL), where models can be heterogeneous. We also add two simple decision-making models to the set of low-level agents, diversifying the pool of available agents and thus increasing overall behaviour flexibility. We perform multiple experiments showing the strengths of a hierarchical approach and test various prediction models at both levels. We also use a risk-based reward at the high level, which transforms the overall problem into a risk-return optimization. This type of reward shows a significant reduction in risk while minimally reducing profits. Overall, the hierarchical approach shows significant promise, especially when the pool of low-level agents is highly diverse.

As the development of high-throughput technologies, more and more high-dimensional or ultra high-dimensional genomic data are generated. Therefore, how to make effective analysis of such data becomes a challenge. Machine learning (ML) algorithms have been widely applied for modelling nonlinear and complicated interactions in a variety of practical fields such as high-dimensional survival data. Recently, the multilayer deep neural network (DNN) models have made remarkable achievements. Thus, a Cox-based DNN prediction survival model (DNNSurv model) , which was built with Keras and Tensorflow, was developed. However, its results were only evaluated to the survival datasets with high-dimensional or large sample sizes. In this paper, we evaluate the prediction performance of the DNNSurv model using ultra high-dimensional and high-dimensional survival datasets, and compare it with three popular ML survival prediction models (i.e., random survival forest and Cox-based LASSO and Ridge models). For this purpose we also present the optimal setting of several hyper-parameters including selection of tuning parameter. The proposed method demonstrates via data analysis that the DNNSurv model performs overall well as compared with the ML models, in terms of three main evaluation measures (i.e., concordance index, time-dependent Brier score and time-dependent AUC) for survival prediction performance.

The traditional standards employed for pain assessment have many limitations. One such limitation is reliability because of inter-observer variability. Therefore, there have been many approaches to automate the task of pain recognition. Recently, deep-learning methods have appeared to solve many challenges, such as feature selection and cases with a small number of data sets. This study provides a systematic review of pain-recognition systems that are based on deep-learning models for the last two years only. Furthermore, it presents the major deep-learning methods that were used in review papers. Finally, it provides a discussion of the challenges and open issues.

A novel coronavirus called 2019-nCoV was recently found in Wuhan, Hubei Province of China, and now is spreading across China and other parts of the world. 2019-nCoV spreads more rapidly than SARS-CoV. Unfortunately, there is no drug to combat the virus. It is of high significance to develop a drug that can combat the virus effectively before the situation gets worse. It usually takes a much longer time to develop a drug using traditional methods. For 2019-nCoV, it is now better to rely on some alternative methods to develop drugs that can combat such a disease effectively since 2019-nCoV is highly homologous to SARS-CoV. In this paper, we first collected virus RNA sequences from the GISAID database, translated the RNA sequences into protein sequences, and built a protein 3D model using homology modeling. Coronavirus main protease is considered to be a major therapeutic target, thus this paper focused on drug screening based on the modeled 2019-nCov_main_protease structure. The deep learning based method DFCNN, developed by our group, can identify/rank the protein-ligand interactions with relatively high accuracy. DFCNN is capable of performing virtual screening quickly since no docking or molecular dynamic simulation is needed. DFCNN identifies potential drugs for 2019-nCoV protease by performing drug screening against 4 chemical compound databases. Also, we performed drug screening for all tripeptides against the binding site of 2019-nCov_main_protease since peptides often show better stability, more bio-availability and negligible immune responses. In the end, we provided the list of possible chemical ligands and peptide drugs for experimental validation.

With the rapid development of Internet and the widely usage of smart devices, massive multimedia data are generated, collected, stored and shared on the Internet. This trend makes cross-modal retrieval problem become a hot issue in this years. Many existing works pay attentions on correlation learning to generate a common subspace for cross-modal correlation measurement, and others uses adversarial learning technique to abate the heterogeneity of multi-modal data. However, very few works combine correlation learning and adversarial learning to bridge the inter-modal semantic gap and diminish cross-modal heterogeneity. This paper propose a novel cross-modal retrieval method, named ALSCOR, which is an end-to-end framework to integrate cross-modal representation learning, correlation learning and adversarial. CCA model, accompanied by two representation model, VisNet and TxtNet is proposed to capture non-linear correlation. Beside, intra-modal classifier and modality classifier are used to learn intra-modal discrimination and minimize the inter-modal heterogeneity. Comprehensive experiments are conducted on three benchmark datasets. The results demonstrate that the proposed ALSCOR has better performance than the state-of-the-arts.

Causality is the most important topic in the history of Western Science, and since the beginning of the statistical paradigm, it meaning has been reconceptualized many times. Causality entered into the realm of multi-causal and statistical scenarios some centuries ago. Despite of widespread critics, today Deep Learning and Machine Learning advances are not weakening causality but are creating a new way of finding indirect factors correlations. This process makes possible us to talk about approximate causality, as well as about a situated causality.

Deep learning has taken over - both in problems beyond the realm of traditional, hand-crafted machine learning paradigms as well as in capturing the imagination of the practitioner sitting on top of petabytes of data. While the public perception about the efficacy of deep neural architectures in complex pattern recognition tasks grows, sequentially up-to-date primers on the current state of affairs must follow. In this review, we seek to present a refresher of the many different stacked, connectionist networks that make up the deep learning architectures followed by automatic architecture optimization protocols using multi-agent approaches. Further, since guaranteeing system uptime is fast becoming an indispensable asset across multiple industrial modalities, we include an investigative section on testing neural networks for fault detection and subsequent mitigation. This is followed by an exploratory survey of several application areas where deep learning has emerged as a game-changing technology - be it anomalous behavior detection in financial applications or financial time-series forecasting, predictive and prescriptive analytics, medical imaging, natural language processing or power systems research. The thrust of this review is on outlining emerging areas of application-oriented research within the deep learning community as well as to provide a handy reference to researchers seeking to embrace deep learning in their work for what it is: statistical pattern recognizers with unparalleled hierarchical structure learning capacity with the ability to scale with information.

Digital images have become an important carrier for people to access information in the information age. However, with the development of the technology, digital images are vulnerable to illegal access and tampering, to the extent that they pose a serious threat to personal privacy, social order and national security. Therefore, image forensic techniques have become an important research topic in the field of multimedia information security. In recent years, deep learning technology has been widely applied in the field of image forensics and the performance achieved has significantly exceeded the conventional forensic algorithms. This survey compares the state-of-the-art image forensic techniques based on deep learning in recent years. The image forensic techniques are divided into passive and active forensics. In passive forensics, forgery detection techniques are reviewed, and the basic framework, evaluation metrics and commonly used datasets for forgery detection are presented. The performance, advantages and disadvantages of existing methods are also compared and analyzed according to different types of detection. In active forensics, robust image watermarking techniques are overviewed, the evaluation metrics and basic framework of robust watermarking techniques are presented. The technical characteristics and performance of existing methods are analyzed based on the different types of attacks on images. Finally, future research directions and conclusions are given to provide useful suggestions for people in image forensics and related research fields.

Deep learning (DL) models have been recently widely used to extract task-oriented patterns from large scale of datasets, and to improve the data image understanding and analysis accuracy in many different decision-making processes for tasks such as image classification, segmentation, detection, and so on. However, in practice, the performances of DL models are easily affected by environmental illumination conditions. Conversely, DL models can also be utilized for extracting the illumination hints from the images, and these hints are critically useful for improving the model robustness, classifying the environmental scenes, estimating scene depth information, and rendering 3D objects. In this study, an extensive and exhaustive review is carried out for DL based color constancy, indoor and outdoor illumination estimation, and image depth estimations with the considerations of strengths and weaknesses of DL models. This study also explores the different network designs and the paradoxes in parameter optimization during the model training. Current technology barriers involved in implementing these models and recommendations to overcome these barriers are also suggested in the review.

Recently, the application of deep reinforcement learning in recommender system is flourishing and stands out by overcoming drawbacks of traditional methods and achieving high recommendation quality. The dynamics, long-term returns and sparse data issues in recommender system have been effectively solved. But the application of deep reinforcement learning brings problems of interpretability, overfitting, complex reward function design, and user cold start. This paper proposed a tag-aware recommender system based on deep reinforcement learning without complex function design, taking advantage of tags to make up for the interpretability problems existing in recommender system. Our experiment is carried out on MovieLens dataset. The result shows that, DRL based recommender system is superior than traditional algorithms in minimum error and the application of tags has little effect on accuracy when making up for interpretability. In addition, DRL based recommender system has excellent performance on user cold start problems.

Properties of spatially explicit data are often ignored or inadequately handled in machine learning for spatial domains of application. At the same time, resources that would identify these properties and investigate their influence and methods to handle them in machine learning applications are lagging behind. In this survey of the literature, we seek to identify and discuss spatial properties of data that influence the performance of machine learning. We review some of the best practices in handling such properties in spatial domains and discuss their advantages and disadvantages. We recognize two broad strands in this literature. In the first, the properties of spatial data are developed in the spatial observation matrix without amending the substance of the learning algorithm; in the other, spatial data properties are handled in the learning algorithm itself. While the latter have been far less explored, we argue they offer the most promising prospects for the future of spatial machine learning.

The study aimed to investigate the impact of incorporating deep learning strategies in mathematics instruction on students' problem-solving abilities, critical thinking skills, academic achievement, and mathematical skills development. The study population consisted of 280 students, with a sample size of 162 participants. Employing a purely quantitative approach, data collection utilized questionnaires and achievement tests as instruments. Data analysis involved simple linear regression, descriptive analysis, and Pearson correlation. The results of the study revealed that there was no significant enhancement observed in students' problem-solving abilities and critical thinking skills when deep learning strategies were incorporated into mathematics instruction. Additionally, students' attitudes towards deep learning activities in mathematics showed no effect on their academic achievement. Furthermore, students' engagement levels in deep learning activities did not impact their development of mathematical skills. Lastly, the frequency of deep learning activities in the mathematics classroom did not show a significant relationship with students' academic performance. These findings suggest that the implementation of deep learning strategies in mathematics instruction did not yield notable improvements in the specified areas based on the study outcomes.

Deep learning heavily relies on statistical correlations to drive artificial intelligence (AI) innovations, particularly in computer vision applications like autonomous driving and robotics. However, despite providing a solid foundation for deep learning, these statistical correlations can be vulnerable to unforeseen and uncontrolled factors. The lack of prior knowledge guidance can result in spurious correlations, introducing confounding factors and affecting the model's robustness. To address this challenge, recent research efforts have focused on integrating causal theory into deep learning methodologies. By modelling the inherent and unbiased causal structure, causal theory can potentially mitigate the impact of spurious correlations effectively. Hence, this paper explores the basics of causal methodologies in image classification.

An accurate, economical, and reliable device for detecting falls in persons ambulating with the assistance of an orthopedic walker is crucially important for the elderly and patients with limited mobility. Existing wearable devices such as wristbands are not designed for walker users, and patients may not wear them at all times. This research proposes a novel idea of attaching an internet-of-things (IoT) device with an inertial measurement unit (IMU) sensor directly to an orthopedic walker to perform real-time fall detection as well as activity logging. A dataset is collected and labeled for walker users in four activities, including idle, motion, step, and fall. Classic machine learning algorithms are evaluated using the dataset by comparing their classification performance. Deep learning with convolutional neural network (CNN) is also explored. Furthermore, the hardware prototype is designed by integrating a low-power microcontroller for onboard machine learning, an IMU sensor, a rechargeable battery, and Bluetooth wireless connectivity. The research results show the promise of improved safety and well-being of walker users.

Agriculture is one of the indispensable fields for the survival of mankind. Potatoes also have a significant role in the field of agriculture. The quality and quantity of potatoes are significantly impacted by several diseases, such as early blight and late blight, and manual interpretation of these leaf diseases is time-consuming and inconvenient. Fortunately, leaf appearance is used to detect diseases in potato plants. Productivity will considerably rise if the infections are detected early. Different types of image processing methods and machine learning methods are used for early recognition of those diseases from leaf images so that the product losses are decreased significantly. For the incredible and marvelous performance of CNN, it is the most popular deep learning method that is used immensely for the recognition of leaf diseases from images. Several pretrained deep learning models, such as VGG16, ResNet50, InceptionV3, MobileNetV2, Xception, and a deep learning model developed using CNN, are employed for potato leaf disease classification and recognition on the same dataset. The transfer learning technique is applied to the pretrained model and the data augmentation technique is applied to the Proposed CNN model for potato disease classification from leaf samples. Compared to pretrained models, the proposed CNN model offers the lowest loss and highest accuracy for potato leaf disease detection while using fewer parameters and layers. It achieves the best performance with a test accuracy of 99.33% compared to other Pretrained models used in the diagnosis of potato leaf disease.

In this paper, we introduce an innovative approach to distinguish Gliosarcoma (GSM) from Glioblastoma (GBM). Our method combines causal fuzzy logic rules with the Big Bird architecture, a Transformer-based Deep Learning algorithm. Unlike prior research, which often relied on statistical models to reduce dataset dimensions before causal analysis, our approach harnesses the complete dataset in tandem with our causal fuzzy Big Bird architecture. Additionally, we benchmark our results not only against previous Gliosarcoma/Glioblastoma studies but also against GPT-2 for a comprehensive evaluation.

Digital Microfluidic Biochips (DMFBs), used in various kinds of fields like DNA analysis, clinical diagnosis, and PCR testing, have made biochemical experiments more compact, efficient, and user-friendly than previous ways. However, their reliability is often compromised by their inability to adapt to all kinds of errors. All errors in biochips can be categorized into two types: known errors and unknown errors. Known errors are detectable before the start of the routing process through sensors or cameras. Unknown errors, in contrast, become apparent only during the routing process and remain undetected by sensors or cameras, which is the biggest issue to unexpectedly stop the routing process and diminishes the reliability of biochips. This paper introduces a deep reinforcement learning-based routing algorithm designed to manage not only known errors but also unknown errors. Our experiments demonstrate that our algorithm outperforms previous ones in terms of the success rate of the routing in the scenario including both known errors and unknown errors. Additionally, our algorithm contributes to detecting unknown errors during the routing process and identifying the most efficient routing path with high probability.

Deep learning machines are computational models composed of multiple processing layers of adaptive weights to learn representations of data with multiple levels of abstraction. Their structure is mainly reflecting the intuitive plausibility of decomposing a problem into multiple levels of computation and representation since it is believed that higher layers of representation allow a system to learn complex functions. Surprisingly, after decades of research, from learning and design perspectives these models are still deployed in a heuristic manner. In this paper, deep learning feed-forward machines are modeled from a statistical mechanics point of view as disordered physical systems where its macroscopic behavior is determined in terms of the interactions defined between the basic constituent of these models, namely, the artificial neuron. They are viewed as the equilibrium states of a theoretical body that is subject to the law of increase of the entropy. The study of the changes in energy of the body when passing from one equilibrium state to another is used to understand the structure and role of the phase space of the system, the stability of the equilibrium states, and the resulting degree of disorder. It is shown that the topology of these models is strongly linked to their stability and resulting level of disorder. Furthermore, the proposed theoretical characterization permit to assess the thermodynamic efficiency with which information can be processed by these models, and to provide a practical methodology to quantitatively estimate and compare their expected learning and generalization capabilities. These theoretical results provides new insights to the theory of deep learning and their implications are shown to be consistent through a set of benchmarks designed to experimentally assess their validity.

Abstract: The main contribution of this paper is to introduce a new iterative training algorithm for restricted Boltzmann machines. The proposed learning path is inspired from online sequential extreme learning machine one of extreme learning machine variants which deals with time accumulated sequences of data with fixed or varied sizes. Recursive least squares rules are integrated for weights adaptation to avoid learning rate tuning and local minimum issues. The proposed approach is compared to one of the well known training algorithms for Boltzmann machines named “contrastive divergence”, in term of time, accuracy and algorithmic complexity under the same conditions. Results strongly encourage the new given rules during data reconstruction.

We propose a deep reinforcement learning based manipulator path tracking method to solve the computationally difficult and non-unique problem of manipulator path tracking methods based on inverse kinematics. By transforming the path tracking task into a sequence decision problem, our method adopts an end-to-end learning method for closed-loop control and avoids the process of finding the inverse solution. We first explored the feasibility of the deep reinforcement learning method in the path tracking of the manipulator. After verifying the feasibility, the path tracking of the multi-degree-of-freedom(multi-DOF) manipulator was realized by combining the maximum entropy deep reinforcement learning algorithm. The experimental results show that our method has a good effect on the path tracking of the manipulator, which not only avoids the process of finding the inverse kinematics solution, but also requires no dynamic model. Therefore, we believe that our method has great significance in the study of manipulator path tracking.

The application of machine learning techniques to satellite imagery has been the subject of interest in recent years. The increase in quality and quantity of images, made available by Earth observation programs, such as the Copernicus program, led to the generation of large amounts of data. Among the various applications of this data is the creation of land cover maps. The present work aimed to create machine learning models capable of accurately segment and classify satellite images, to automatically generate a land cover map of the Portuguese territory. Several experiments were carried out with the spectral bands of the Sentinel-2 satellite, with vegetation indices, and with several sets of land cover classes. Three machine learning architectures were evaluated, which adopt two different techniques for image classification. One of the classification techniques follows an object-oriented approach, and in this case the architecture adopted in our models was a U-Net artificial neural network. The other classification technique is pixel-oriented, and the machine learning models tested were random forest and support vector machine. The overall accuracy of the results obtained ranged from 68.6% to 94.75%, depending strongly on the number of classes into which the land cover is classified. The result of 94.75% was obtained when classifying the land cover only into 5 classes. However, a very interesting accuracy of 92.37% was achieved by the model when trained to classify 8 classes. These results are superior to those reported in the related bibliography.

Over a billion people live in slums, with poor sanitation, education, property rights and working conditions having direct impact on current residents and future generations. A key problem in relation to slums is slum mapping. Without delineations of where all slum settlements are, informed decisions cannot be made by policymakers in order to benefit the most in need. Satellite images have been used in combination with machine learning models to try and fill the gap in data availability of slum locations. Deep learning has been used on RGB images with some success but since labeled satellite images of slums are relatively low quality and the physical/visual manifestation of slums significantly varies within and across countries, it is important to quantify the uncertainty of predictions for reliable application in downstream tasks. Our solution is to train Monte Carlo dropout U-Net models on multispectral 13-band Sentinel-2 images from which we can calculate pixelwise epistemic (model) and aleatoric (data) uncertainty in our predictions. We trained our model on labelled images of Mumbai and verified our epistemic and aleatoric uncertainty quantification approach using altered models trained on modified datasets. We also used SHAP values to investigate how the different features contribute towards the model’s predictions and this showed that certain short-wave infrared and red-edge image bands are powerful features for determining the locations of slums within images. Having created our model with uncertainty quantification, in the future it can be applied to downstream tasks and decision-makers will know where predictions have been made with low uncertainty, giving them greater confidence in its deployment.

Layer-wise 3D surface morphology information is critical for the quality monitoring and control of additive manufacturing (AM) processes. However, most of the existing 3D scan technologies are either contact or time consuming, which are not capable of obtaining the 3D surface morphology data in a real-time manner during the process. Therefore, the objective of this study is to achieve real-time 3D surface data acquisition in AM, which is achieved by a supervised deep learning-based image analysis approach. The key idea of this proposed method is to capture the correlation between 2D image and 3D point cloud, and then quantify this relationship by using a deep learning algorithm, namely, convolutional neural network (CNN). To validate the effectiveness and efficiency of the proposed method, both simulation and real-world case studies were performed. The results demonstrate that this method has strong potential to be applied for real-time surface morphology measurement in AM, as well as other advanced manufacturing processes.

Artificial intelligence methods and application have recently shown great contribution in modeling and prediction of the hydrological processes, climate change, and earth systems. Among them, deep learning and machine learning methods mainly have reported being essential for achieving higher accuracy, robustness, efficiency, computation cost, and overall model performance. This paper presents the state of the art of machine learning and deep learning methods and applications in this realm and the current state, and future trends are discussed. The survey of the advances in machine learning and deep learning are presented through a novel classification of methods. The paper concludes that deep learning is still in the first stages of development, and the research is still progressing. On the other hand, machine learning methods are already established in the fields, and novel methods with higher performance are emerging through ensemble techniques and hybridization.

With the continuous development of network technology, complex network systems generate massive unbalanced attack traffic. Due to the severe imbalance in the quantities of normal samples and attack samples, as well as among different types of attack samples, intrusion detection systems suffer from low detection rates for rare class attack data. In this paper, we propose a geometric synthetic minority oversampling technique based on optimized kernel density estimation algorithm. This method can generate diverse rare class attack data by learning the distribution of rare class attack data while maintaining similarity with the original sample features. Meanwhile, the balanced data is input to a feature extraction module built upon multiple denoising autoencoders, reducing information redundancy in high-dimensional data and improving the detection performance for unknown attacks. Subsequently, a soft voting ensemble learning technique is utilized for multi-class anomaly detection on the balanced and dimensionally reduced data. Finally, an intrusion detection system is constructed based on data preprocessing, imbalance handling, feature extraction, and anomaly detection modules, and validated on the NSL-KDD and N-BaIoT datasets. Comparative experiments with baseline models and other state-of-the-art methods demonstrate that the proposed system improves the detection rate of rare class attack data. Furthermore, it achieves a good overall detection rate on the Internet of Things dataset (N-BaIoT), indicating its strong applicability.

People localization is a key building block in many applications. In this paper, we propose a deep learning based approach that significantly improves the localization accuracy and reduces the runtime of Wi-Fi based localization systems. Three variants of the deep learning approach are proposed, a sub-task architecture, an end-to-end architecture, and an architecture that incorporates prior knowledge. The performance of the three architectures under different conditions is evaluated and the significant improvement of the three architectures over existing approaches is demonstrated.

Tuning tensor program generation involves navigating a vast search space to find optimal program transformations and measurements for a program on target hardware. The complexity of this process is further amplified by the exponential combinations of transformations, especially in heterogeneous environments. This research addresses these challenges by introducing a novel approach that learns the joint neural network and hardware features space, facilitating knowledge transfer to new, unseen target hardware. A comprehensive analysis is conducted on the existing state-of-the-art dataset, TenSet, including a thorough examination of test split strategies and the proposal of methodologies for dataset pruning. Leveraging an attention-inspired technique, we tailor the tuning of tensor programs to embed both neural network and hardware-specific features. Notably, our approach substantially reduces the dataset size by up to 53% compared to the baseline without compromising Pairwise Comparison Accuracy (PCA). Furthermore, our proposed methodology demonstrates competitive or improved mean inference times with only 25%40% of the baseline tuning time across various networks and target hardware. The attention-based tuner can effectively utilize schedules learned from previous hardware program measurements to optimize tensor program tuning on previously unseen hardware, achieving a top-5 accuracy exceeding 90%. This research introduces a significant advancement in auto-tuning tensor program generation, addressing the complexities associated with heterogeneous environments and showcasing promising results regarding efficiency and accuracy.

In August 2020, the World Health Assembly launched a global initiative to eliminate cervical cancer by 2030, setting three primary targets. One key goal is to achieve a 70% screening coverage rate for cervical cancer, primarily relying on the precise analysis of Papanicolaou (Pap) or digital Pap smears. However, the responsibility of reviewing Pap smear samples to identify potentially cancerous cells primarily falls on pathologists, a task known to be exceptionally challenging and time-consuming. This paper proposes a solution to address the shortage of pathologists for cervical cancer screening. It leverages the OpenAI-GYM API to create a deep reinforcement learning environment utilizing liquid-based Pap smear images. By employing the Proximal Policy Optimization algorithm, autonomous agents navigate Pap smear images, identifying cells with the aid of rewards, penalties, and accumulated experiences. Furthermore, the use of a pre-trained convolutional neuronal network like Res-Net54 enhances the classification of detected cells based on their potential for malignancy. The ultimate goal of this study is to develop a highly efficient, automated Papanicolaou analysis system, ultimately reducing the need for human intervention in regions with limited pathologists.

Image registration (IR) is a process that deforms images to align them with respect to a reference space, making it easier for medical practitioners to examine various medical images in a standardized reference frame, such as having the same rotation and scale. This document introduces image registration using a simple numeric example. It provides a definition of image registration along with a space-oriented symbolic representation. This review covers various aspects of image transformations, including affine, deformable, invertible, and bidirectional transformations, as well as medical image registration algorithms such as Voxelmorph, Demons, SyN, Iterative Closest Point, and SynthMorph. It also explores atlas-based registration and multistage image registration techniques, including coarse-fine and pyramid approaches. Furthermore, this survey paper discusses medical image registration taxonomies, datasets, evaluation measures, such as correlation-based metrics, segmentation-based metrics, processing time, and model size. It also explores applications in image-guided surgery, motion tracking, and tumor diagnosis. Finally, the document addresses future research directions, including the further development of transformers.

Physical inactivity increases the risk of many adverse health conditions, including the world’s major non-communicable diseases, such as coronary heart disease, type 2 diabetes, and breast and colon cancers, shortening life expectancy. There are minimal medical care and personal trainers’ methods to monitor a patient’s actual physical activity types. To improve activity monitoring, we propose an artificial-intelligence-based approach to classify the physical movement activity patterns. In more detail, we employ two deep learning (DL) methods, namely a deep feed-forward neural network (DNN) and a deep recurrent neural network (RNN) for this purpose. We evaluate the proposed models on two physical movement datasets collected from several volunteers who carried tri-axial accelerometer sensors. The first dataset is from the UCI machine learning repository, which contains 14 different activities-of-daily-life (ADL) and is collected from 16 volunteers who carried a single wrist-worn tri-axial accelerometer. The second dataset includes ten other ADLs and is gathered from 8 volunteers who placed the sensors on their hips. Our experiment results show that the RNN model provides the accuracy performance compared to the state-of-the-art methods in classifying the fundamental movement patterns with an overall accuracy of 84.89% and an overall F1-score of 82.56%. Our results indicate that the proposed method will provide the medical doctors and trainers a promising way to precisely track and understand a patient’s physical activities for better treatment.

The Ebola virus disease (EVD) is an extremely contagious and fatal illness caused by the Ebola virus. Recently, Uganda witnessed an outbreak of EVD, which generated a lot of attention on various social media platforms. To ensure effective communication and implementation of targeted health interventions, it is crucial for stakeholders to comprehend the sentiments expressed in the posts and discussions on these online platforms. In this study, we used deep learning techniques to analyze the sentiments expressed in Ebola-related tweets during the outbreak. We explored the application of three deep learning techniques to classify the sentiments in 8,395 tweets as positive, neutral, or negative. The techniques examined included a 6-layer convolutional neural network (CNN), a 6-layer long-short-term memory model (LSTM), and an 8-layer Bidirectional Encoder Representations from Transformers (BERT) model. The study found that the BERT model outperformed both the CNN and LSTM-based models across all the evaluation metrics, achieving a remarkable classification accuracy of 95\%. These findings confirm the reported effectiveness of transformer-based architectures in tasks related to natural language processing, such as sentiment analysis.

Chlorophyll-a (Chla) is a crucial pigment in phytoplankton, playing a vital role in determining phytoplankton biomass and water nutrient status. However, in optically complex water bodies, Chla concentration is no longer the primary factor influencing remote sensing spectral reflectance signals, leading to significant errors in traditional Chla concentration estimation methods. With advancements in in-situ measurements, synchronized satellite data, and computer technology, machine learning algorithms have become popular in Chla concentration retrieval. Nevertheless, when using machine learning methods to estimate Chla concentration, abrupt changes in Chla values can disrupt the spatiotemporal smoothness of the retrieval results. Therefore, this study proposes a two-stage approach to enhance the accuracy of Chla concentration estimation in optically complex water bodies. In the first stage, a one-dimensional convolutional neural network (1DCNN) is employed for precise Chla retrieval, and in the second stage, the regression layer of the 1DCNN is replaced with Support Vector Regression (SVR). The research findings are as follows: (1) In the first stage, the performance metrics (R², RMSE, RMLSE, Bias, MAE) of the 1DCNN outperform state-of-the-art algorithms (OCI, SVR, RFR) on the test dataset. (2) After the second stage, the performance further improves, with the metrics achieving values of 0.892, 11.243, 0.052, 1.056, and 1.444, respectively. (3) In mid-to-high latitude regions, the inversion performance of 1DCNN\SVR is superior to other algorithms, exhibiting richer details and higher noise tolerance in nearshore areas. (4) 1DCNN\SVR demonstrates high inversion capabilities in water bodies with medium to high nutrient levels.

In this brief letter we report our initial results on the application of deep-learning to the massive MIMO channel estimation challenge. We show that it is possible to estimate wireless channels and that the possibility of mitigating pilot-contamination with deep-learning is possible given that the leaning model underwent an extensive training-phase and that it has been presented with a large number of different channel conditions.

This paper presents a two-stage hierarchical neural network using image classification and object detection algorithms as key building blocks for a system that automatically detects a potential design right infringement. This neural network is trained to return the Top-N original design right records that highly resemble the input image of a counterfeit. Design rights specify the unique aesthetic characteristics of a product. Due to the rapid change of trends, new design rights are continuously generated. This work proposes an Ensemble Neural Network (ENN), an artificial neural network model that aims to deal with a large amount of counterfeit data and design right records that are frequently added and deleted. At first, we performed image classification and objection detection learning per design right using the existing models with a proven track record of high accuracy. The distributed models form the backbone of the ENN and yield intermediate results aggregated at a master neural network. This master neural network is a deep residual network paired with a fully connected network. This ensemble layer is trained to determine the sub-models that return the best result for a given input image of a product. In the final stage, the ENN model multiples the inferred similarity coefficients to the weighted input vectors produced by the individual sub-models to assess the similarity between the test input image and the existing product design rights to see any sign of violation. Given 84 design rights and the sample product images taken meticulously under various conditions, our ENN model achieved average Top-1 and Top-3 accuracies of 98.409% and 99.460%, respectively. Upon introducing new design rights data, a partial update of the inference model was done an order of magnitude faster than the single model. ENN maintained a high level of accuracy as it scaled out to handle more design rights. Therefore, the ENN model is expected to offer practical help to the inspectors in the field, such as the customs at the border that deal with a swarm of products.

Building deep learning models proposed by third parties can become a simple task when specialized libraries are used. However, a great deal of mystery still surrounds the design of new models or the modification of existing ones. These tasks require in-depth knowledge of the different components or building blocks as well as their dimensions. This information is limited and broken up in different literature. In this article, we collect and explain in-depth the building blocks to design deep learning models, starting from the artificial neuron to the concepts involved in building deep neural networks. Furthermore, the implementation of each building block is exemplified using the Keras library.

In training a deep learning system to perform audio transcription, two practical problems may arise. Firstly, most datasets are weakly labelled, having only a list of events present in each recording without any temporal information for training. Secondly, deep neural networks need a very large amount of labelled training data to achieve good quality performance, yet in practice it is difficult to collect enough samples for most classes of interest. In this paper, we propose factorising the final task of audio transcription into multiple intermediate tasks in order to improve the training performance when dealing with this kind of low-resource datasets. We evaluate three data-efficient approaches of training a stacked convolutional and recurrent neural network for the intermediate tasks. Our results show that different methods of training have different advantages and disadvantages.

Our daily activities hinge on the flexibility of our fingers, and a fractured finger can significantly disrupt these routines. The finger bones enable us to bend and fold the fingers and thumb to pick up or grasp objects and do all of our daily activities. A broken finger can cause adverse effects on our daily life activities. It is important to treat broken finger as soon as possible. Swift and precise treatment begins with capturing multiple X-rays, followed by the critical step of fracture detection in these images. Relying on the naked eye for this task carries the risk of overlooking small fractures. To address this issue an automated diagnoses of fractured fingers from images is required for which the current research employs advanced deep learning models—ResNet34, ResNet50, ResNet101, ResNet152, VGG-16, and VGG-19—to classify finger images from the Musculoskeletal Radiographs (MURA) dataset as either fractured or non-fractured. The results emphasize the consistently strong performance of ResNet models, attaining an impressive accuracy of 81.9%. This surpasses VGG models by 3.4% and establishes ResNet as the new benchmark for state-of-the-art accuracy.

A continuous Deep Belief Network (cDBN) with two hidden layers is proposed in this paper, focusing on the problem of weak feature learning ability when dealing with continuous data. In cDBN, the input data is trained in an unsupervised way by using continuous version of transfer functions, the contrastive divergence is designed in hidden layer training process to raise convergence speed, an improved dropout strategy is then implemented in unsupervised training to realize features learning by de-cooperating between the units, and then the network is fine-tuned using back propagation algorithm. Besides, hyper-parameters are analysed through stability analysis to assure the network can find the optimal. Finally, the experiments on Lorenz chaos series, CATS benchmark and other real world like CO2 and waste water parameters forecasting show that cDBN has the advantage of higher accuracy, simpler structure and faster convergence speed than other methods.

Continuous greenhouse gas emissions are causing global warming and impacting the habitats of many animals. Researchers in the field of electric power are making efforts to mitigate this situation. Operating and maintaining the power grid in an economic, low-carbon, and stable is challenging. To address the issue, we propose a grid dispatching technique that combines prediction technology, reinforcement learning, and optimization technology. Prediction technology can forecast future power demand and solar power generation, while reinforcement learning and optimization technology can make charging and discharging decisions for energy storage devices based on current and future grid conditions. In the power system, the aggregation of distributed energy resources increases uncertainty, particularly due to the fluctuating generation of renewable energy. This requires the use of advanced predictive control techniques to ensure long-term economic and decarbonization goals. In this paper, we present a real-time dispatching framework that integrates deep learning-based prediction, reinforcement learning-based decision-making, and stochastic optimization techniques. The framework can rapidly adapt to target uncertainty caused by various factors in real-time data distribution and control processes. The proposed framework achieved global Champion in the NeurIPS Challenge 2022 competition and demonstrated its effectiveness in practical scenarios of intelligent building energy management.

The research of object classification and part segmentation is a hot topic in computer vision. A considerable number of studies have been carried out about deep learning on 3D point clouds. However,it is challenging to achieve effective feature learning due to sparsity of point clouds. Recently, a variety of Transformers have been adopted to improve point cloud processing and display great potential. Nevertheless, large numbers of Transformer layers tend to incur huge computational and memory costs. PointNet++ is one of the most influential neural architectures for point cloud understanding. Although the accuracy of PointNet++ has been largely surpassed by recent networks, this does not mean that PointNet++ has no potential.Thus, this paper offer two major contributions that significantly improve PointNet++ performance. Firstly, we introduce a novel contextual feature extraction (CFE) block that significantly enhances the feature extraction capabilities of PointNet++ networks. Secondly, to further enhance feature fusion, we seamlessly integrate a mutual learning (ML) block into the network architecture. By embedding these two innovative blocks within each layer of the network, we not only enrich the network's functionality but also impart it with greater robustness and adaptability. The specific experiments were conducted on the S3DIS (6-fold cross-validation) and Modelnet40 datasets with 86.5% and 92.7% accuracy, respectively, which proved that our model is comparable or even better than most existing methods for classification and segmentation tasks.

Scaffoldings play a critical role as temporary structures in supporting construction processes. Accidents at construction sites frequently stem from issues related to scaffoldings, including insufficient support caused by deviations from the construction design, insecure rod connections, or absence of cross bracing, which result in uneven loading and potential collapse, leading to casualties. In this study, we introduce a deep-learning-based, augmented reality (AR)-enabled system called scaffolding assembly deficiency detection system (SADDS) to assist field inspectors in identifying deficiencies in scaffolding assemblies. Inspectors have the flexibility to utilize SADDS through various devices, such as video cameras, mobile phones, or AR goggles (i.e., Microsoft HoloLens 2), for automated detection of deficiencies in scaffolding assemblies. The training test yielded satisfactory results, with a mean average precision of 0.89 and individual precision values of 0.96, 0.82, 0.90, and 0.89 for “qualified” frames and frames with the “missing cross-tie rod,” “missing lower-tie rod,” and “missing footboard” deficiencies, respectively. Field tests conducted at two construction sites demonstrated improved system performance compared to that in the training test. However, these field tests also revealed certain limitations of the SADDS system.

In recent years, applying deep learning to solve physics problems has attracted much attention. Data-driven deep learning methods produce operators that can learn solutions to the whole system of partial differential equations. However, the existing methods are only evaluated on simple flow equations (e.g., Burger's equation), and only consider the generalization ability on different initial conditions. In this paper, we construct CFDBench, a benchmark with four classic problems in computational fluid dynamics (CFD): lid-driven cavity flow, laminar boundary layer flow in circular tubes, dam flows through the steps, and periodic Karman vortex street. Each flow problem includes data with various boundary conditions, fluid physical properties, and domain geometries. Compared to existing datasets, the advantages of CFDBench are (1) comprehensive. It contains common physical parameters such as velocity, pressure, and cavity fraction. (2) realistic. It is very suitable for deep learning solutions of fluid mechanics equations. (3) challenging. It has a certain learning difficulty, prompting to find models with strong learning ability. (4) standardized. CFDBench facilitates a comprehensive and fair comparison of different deep learning methods for CFD. We make appropriate modifications to popular deep neural networks to apply them to CFDBench and enable the accommodation of more changing inputs. The evaluation on CFDBench reveals some new shortcomings of existing works and we propose possible directions for solving such problems.

Depression is considered by WHO as the main contributor to global disability and it poses dangerous threats to approximately all aspects of human life, in particular public and private health. This mental disorder is usually characterized by considerable changes in feelings, routines, or thoughts. With respect to the fact that early diagnosis of this illness would be of critical importance ineffective treatment, some development has occurred in the purpose of depression detection. EEG signals reflect the working status of the human brain by which are considered the most proper tools for a depression diagnosis. Deep learning algorithms have the capacity of pattern discovery and extracting features from the raw data which is fed into them. Owing to this significant characteristic of deep learning, recently, these methods have intensely utilized in the diverse field of researches, specifically medicine and healthcare. Thereby, in this article, we aimed to review all papers concentrated on using deep learning to detect or predict depressive subjects with the help of EEG signals as input data. Regarding the adopted search method, we finally evaluated 22 articles between 2016 and 2021. This article which is organized according to the systematic literature review (SLR) method, provides complete summaries of all exploited studies and compares the noticeable aspects of them. Moreover, some statistical analysis performs to gain a depth perception of the general ideas of the latest researches in this area. A pattern of a five-step procedure was also established by which almost all reviewed articles fulfilled the goal of depression detection. Finally, open issues and challenges in this way of depression diagnosis or prediction and suggested works as the future directions discussed.

In the recent years, intelligent data-driven faultdiagnosis methods on gearboxes have been successfully developedand popularly applied in the industries. Currently, most ofthe machine learning techniques require that the training andtesting data are from the same distribution. However, thisassumption is difficult to be met in the real industries, sincethe gearbox operating conditions usually change in practice,which results in significant data distribution gap and diagnosticperformance deteriorations in applying the learned knowledgeon the new conditions. This paper proposes a deep learning-based domain adaptation method to address this issue. Theraw current signals are directly used as the model inputs fordiagnostics, which are easy to collect in the real industries andfacilitate practical applications. The maximum mean discrepancymetric is introduced to the deep neural network, the optimizationof which guarantees the extraction of generalized machineryhealth condition features across different operating conditions.The experiments on a real-world gearbox condition monitoringdataset validate the effectiveness of the proposed method, whichoffers a promising tool for cross-domain diagnosis in the realindustries.

This paper investigates the use of deep learning techniques to perform energy demand forecasting. Specifically, the authors have adapted a deep neural network originally thought for image classification and composed of a convolutional neural network (CNN) followed by a multilayered fully connected artificial neural network (ANN). The convolutional part of the network was fed with a grid of temperature forecasting data distributed in the area of interest in order to extract a featured temperature. The subsequent ANN is then fed with this calculated temperature along with other data related to the timing of the forecast. The proposed structure was first trained and then used in a real setting aimed to provide the French energy demand forecast using ARPEGE forecasting weather data. The results show that the performance of this approach is in the line of the performance provided by the reference RTE subscription-based service, which opens the possibility to obtain high accuracy forecasting using widely accessible deep learning techniques through open-source machine learning platforms.

Despite significant advances in tumor biology and clinical therapeutics, metastasis remains the primary cause of cancer-related deaths. While RNA-seq technology has been used extensively to study metastatic cancer characteristics, challenges persist in acquiring adequate transcriptomic data. To overcome this challenge, we propose MetGen, a generative contrastive learning based on deep learning model. MetGen generates synthetic metastatic cancer expression profiles using primary cancer and normal tissue expression data. Our results demonstrate that MetGen generates comparable samples to actual metastatic cancer samples, and we discuss the learning mechanism of MetGen. Additionally, we demonstrate MetGen's interpretability using metastatic prostate cancer and metastatic breast cancer. MetGen has learned highly relevant signatures in cancer, tissue, and tumor microenvironment, such as immune response and the metastasis process, which potentially fosters a more comprehensive understanding of metastatic cancer biology. The development of MetGen represents a significant step toward the study of metastatic cancer biology by providing a generative model that identifies candidate therapeutic targets for the treatment of metastatic cancer.

Time series (TS)-based predictions are made by examining the behaviour of historical data to forecast future values. Multiple industries; such as stock market trading, power load forecasting, medical monitoring, and intrusion detection; frequently use it. Several variables; such as the performance of other markets as well as the economic situation of a country which affect its market performance; significantly affect the prediction of stock market prices. Therefore, this present study uses numerous variables; such as the opening, lowest, highest, and closing prices; to predict the indices of the stock market of the Kingdom of Saudi Arabia (KSA). Successfully accomplishing an investment goal largely depends on choosing the right stocks to buy, sell, or maintain. The project's output is the projected closing prices (regression) over the next seven days, which helps investors make the best decisions. This present study used exponential smoothing (ES) to remove noise from the input data obtained from the Saudi Stock Exchange, or Tadawul, before using a multivariate long short-term memory (LSTM) deep learning (DL) algorithm to forecast stock market prices. The proposed multivariate LSTMDL model had satisfactory prediction rates of 97.49% and 92.19% for the univariate model. Therefore, it can be used to effectively predict stock market prices. The results also indicate that DL as well as multiple sources of information can be used to accurately predict stock markets.

MicroRNAs (miRNAs) are small non-coding RNAs that play a central role in the post-transcriptional regulation of biological processes. miRNAs regulate transcripts by direct binding involving the Argonaute protein family. The exact rules of binding are not known, and several in silico miRNA target prediction methods have been developed to date. Deep Learning has recently revolutionized miRNA target prediction. However, the higher predictive power comes with decreased ability to interpret increasingly complex models. Here, we present a novel interpretation technique, called attribution sequence alignment, for miRNA target site prediction models that can interpret such Deep Learning models on a two-dimensional representation of miRNA and putative target sequence. Our method produces a human readable visual representation of miRNA:target interactions and can be used as a proxy for further interpretation of biological concepts learned by the neural network. We demonstrate applications of this method in clustering of experimental data into binding classes, as well as using the method to narrow down predicted miRNA binding sites on long transcript sequences. Importantly, the presented method works with any neural network model trained on a two-dimensional representation of interactions and can be easily extended to further domains such as protein-protein interactions.

COVID-19 has a severe risk of spreading rapidly, the quick identification of which is essential. In this regard, chest radiology images have proven to be a practical screening approach for COVID-19 affected patients. This study proposes a deep learning-based approach using Densenet-121 to detect COVID-19 patients effectively. We have trained and tested our model on the COVIDx dataset and performed both 2-class and 3-class classification, achieving 96.49% and 93.71% accuracy, respectively. By successfully utilizing transfer learning, we achieve comparable performance to the state-of-the-art method while using 15x fewer model parameters. Moreover, we performed an interpretability analysis using Grad-CAM to highlight the most significant image regions at test time. Finally, we developed a website that takes chest radiology images as input and detects the presence of COVID-19 or pneumonia and a heatmap highlighting the infected regions. Source code for reproducing results and model weights are available.

Skin cancer is one of the widespread diseases that typically develop on the skin due to continuous exposure to sunlight. Although cancer can appear on any part of the human body, skin cancer reports account for over half of all cancer occurrences worldwide. There are substantial obstacles to the precise diagnosis and classification of skin lesions because of the morphological variety and indistinguishable characteristics across skin malignancies. Recently, Deep Learning models have been used in the field of image-based lesion diagnosis, and it has demonstrated diagnostic efficiency on par with that of dermatologists. To increase classification efficiency and accuracy for skin lesions, a cutting-edge multi-layer deep Convolutional Neural Network (CNN) termed SkinLesNet has been built in this study. The ResNetV50 and VGG16 models have been carefully compared to review the performance of the proposed model. The dataset used in this study, PAD-UFES-20, contains 1314 samples in total and includes three common forms of skin lesions. The proposed approach, SkinLesNet, significantly outperforms the well-known compared models in the given conditions

Barrett's esopaghagus (BE) is a known precursor of esophageal adenocarcinoma (EAC). Patients with BE undergo regular surveillance to early detect stages of EAC. Volumetric laser endomicroscopy (VLE) is a novel technology capable of imaging the inner tissue layers of the esophagus over a 6-cm length scan. However, interpretation of full VLE scans is still a challenge for human observers. In this work, we train an ensemble of deep convolutional neural networks to detect neoplasia in BE patients, using a dataset of images acquired with VLE in a multicenter study. We achieve values of AUC=$0.96$ on the unseen test dataset and we compare our results with previous work done with VLE analysis. Our method for detecting neoplasia in BE patients facilitates future advances on patient treatment and provides clinicians with new assisting solutions to process and better understand VLE data.

Deep learning has already revolutionised the way we process a wide range of data, in many areas of our daily life. The ability to learn abstractions and relationships from heterogeneous data, has provided impressively accurate prediction and classification tools to handle increasingly big datasets. This has a significant impact on the growing wealth of omics datasets, with the unprecedented opportunity for a better understanding of the complexity of living organisms. While this revolution is transforming the way we analyse these data, explainable deep learning is emerging as an additional tool with the potential to change the way we interpret biological data. Explainability addresses critical issues such as transparency, so important when computational tools are introduced especially in clinical environments. Moreover, it empowers artificial intelligence with the capability to provide new insights in the input data, thus adding an element of discovery to these already powerful resources. In this review we provide an overview of the transformative effects explainable deep learning is having on multiple sectors, ranging from genome engineering and genomics, from radiomics to drug design and clinical trials. We offer a perspective to life scientists, to better understand the potential of these tools, and a motivation to implement them in their research, by suggesting learning resources they can use to move their first steps in this field.

Deep learning techniques, such as convolutional neural networks (CNN), generative adversarial networks (GAN), and graph neural networks (GNN), have over the past decade changed the ac-curacy of prediction in many diverse fields. In recent years, the application of deep learning tech-niques in computer vision tasks in pathology demonstrated extraordinary potential in assisting clinicians, automating diagnosis, and reducing costs for patients. Formerly unknown pathologi-cal evidence, such as morphological features related to specific biomarkers, copy number varia-tions, and other molecular features, were also able to be captured by deep learning models. In this paper, we review popular deep learning methods and some recent publications about their appli-cations in pathology.