While thunderstorms can pose severe risks to property and life, forecasting remains challenging, even at short lead times, as these often arise in meta-stable atmospheric conditions. In this paper, we examine the question of how well we could perform short-term (up to 180min) forecasts using exclusively multi-spectral satellite images and past lighting events as data. We employ representation learning based on deep convolutional neural networks in an “end-to-end” fashion. Here, a crucial problem is handling the imbalance of the positive and negative classes appropriately in order to be able to obtain predictive results (which is not addressed by many previous machine-learning-based approaches). The resulting network outperforms previous methods based on physically-based features and optical flow methods (similar to operational prediction models) and generalizes across different years. A closer examination of the classifier performance over time and under masking of input data indicates that the learned model actually draws most information from structures in the visible spectrum, with infrared imaging sustaining some classification performance during the night.

Clients are increasingly looking for fast and effective means to quickly and frequently survey and communicate the condition of their buildings so that essential repairs and maintenance work can be done in a proactive and timely manner before it becomes too dangerous and expensive. Traditional methods for this type of work commonly comprise of engaging building surveyors to undertake a condition assessment which involves a lengthy site inspection to produce a systematic recording of the physical condition of the building elements, including cost estimates of immediate and projected long-term costs of renewal, repair and maintenance of the building. Current asset condition assessment procedures are extensively time consuming, laborious, and expensive and pose health and safety threats to surveyors, particularly at height and roof levels which are difficult to access. We propose a method for automated detection and localisation of key building defects from images using deep learning and convolution neural networks. The proposed model is based on a pre-trained VGG-16 classifier with Class Activation Mapping (CAM) for object localisation. The model has proven to be robust and able to accurately detect and localise mould growth, stains, and paint deterioration defects arising from dampness in buildings. The approach is being developed with potentials to scale-up to support automated detection of defects and deterioration of buildings in real-time using mobile devices and drones.

Nowadays, technological advancement has transformed traditional vehicles into Au-tonomous Vehicles (A.V.s). In addition, in our daily lives, A.V.s play an important role since they are considered an essential component of smart cities. A.V. is an intelligent vehicle capable of main-taining safe driving by avoiding crashes caused by drivers. Unlike traditional vehicles, which are fully controlled and operated by humans, A.V.s collect information about the outside environment using sensors to ensure safe navigation. Furthermore, A.V.s reduce environmental impact because they usually use electricity to operate instead of fossil fuel, thus decreasing the greenhouse gasses. However, A.V.s could be threatened by cyberattacks, posing risks to human life. For example, re-searchers reported that Wi-Fi technology could be vulnerable to cyberattacks through Tesla and BMW AVs. Therefore, more research is needed to detect cyberattacks targeting the components of A.V.s to mitigate their negative consequences. This research will contribute to the security of A.V.s by detecting cyberattacks at the early stages. First, we inject False Data Injection (FDI) attacks into an A.V. simulation-based system developed by MathWorks. Inc. Second, we collect the dataset generated from the simulation model after integrating the cyberattack. Third, we implement an intelligent symmetrical anomaly detection method to identify FDI attacks targeting the control system of the A.V. through a compromised sensor. We use long short-term memory (LSTM) deep networks to detect FDI attacks in the early stage to ensure the stability of the operation of A.V.s. Our method classifies the collected dataset into two classifications: normal and anomaly data. The ex-perimental result shows that our proposed model's accuracy is 99.95%. To this end, the proposed model outperforms other state-of-the-art models in the same study area.

The spread of the Coronavirus pandemic has been accompanied by an infodemic. The false information that is embedded in the infodemic affects people’s ability to have access to safety information and follow proper procedures to mitigate the risks. This research aims to target the falsehood part of the infodemic, which prominently proliferates in news articles and false medical publications. Here, we present NeoNet, a novel supervised machine learning text mining algorithm that analyzes the content of a document (news article, a medical publication) and assigns a label to it. The algorithm is trained by TFIDF bigram features which contribute a network training model. The algorithm is tested on two different real-world datasets from the CBC news network and Covid-19 publications. In five different fold comparisons, the algorithm predicted a label of an article with a precision of 97-99 %. When compared with prominent algorithms such as Neural Networks, SVM, and Random Forests NeoNet surpassed them. The analysis highlighted the promise of NeoNet in detecting disputed online contents which may contribute negatively to the COVID-19 pandemic.

This research discusses the development of a software-centered EdTech (Education Technology) class model through the implementation of a course titled "3D Time Machine" at a 4-year university in Republic of Korea over two semesters. The course focused on teaching 3D modeling software, Blender, within the history department. The primary objective of offering this course was to equip students from the digital generation with the capability to manipulate digital technology effectively. Through studying historical materials and accumulating domain knowledge, students could construct their narratives from their unique perspectives. This aimed to foster their proficiency in digital technology operation, preparing them for an education environment increasingly centered around virtual worlds. As the use of virtual worlds gains prominence in educational settings, there is a growing need to incorporate curricula that prepare students to thrive as well-prepared citizens in future societies. Assessing the digital competencies of contemporary university students and designing instructional models with particular attention to their needs is becoming increasingly important. This research draws insights from interviews, conducted in both face-to-face and written formats, with students who participated in the "3D Time Machine" course. The interviews revealed valuable insights that can be actively incorporated into the development of software-centered EdTech instructional models. They reported that they expanded their perceptions as they learned how to make their ideas tangible. The course helped students overcome their fear of digital technology. In conclusion, the course effectively employs 3D modeling technology, a software-centered EdTech, as a core element in helping students develop their narratives rapidly and diversely, thereby playing a crucial role in their ability to articulate their unique perspectives.

The present work evaluates the food effect on the absorption of rivaroxaban (Riva), a BCS II drug, from the orally administered commercial immediate-release tablet (Xarelto IR) using physiologically based pharmacokinetic (PBPK) and conventional in vitro- in vivo correlation (IVIVC) models. The bioavailability of Riva upon oral administration of Xarelto IR tablet is reported to exhibit a positive food effect. The PBPK model for Riva was developed and verified using the previously reported in vivo data for oral solution (5 and 10 mg) and Xarelto IR tablet (5 and 10 mg dose strength). Once the PBPK model was established, the in vivo performance of the tablet formulation with the higher dose strength (Xarelto IR tablet 20 mg in fasted and fed state) was predicted using the experimentally obtained data of in vitro permeability, biorelevant solubility and in vitro dynamic dissolution data using United States Pharmacopeia (USP) IV flow-through cell apparatus. In addition, the mathematical IVIVC model was developed using the in vitro dissolution and in vivo profile of 20 mg strength Xarelto IR tablet in fasted condition. Using the developed IVIVC model, the pharmacokinetic (PK) profile of the Xarelto IR tablet in fed condition was predicted and compared with the PK parameters obtained via the PBPK model. A virtual in vivo PK study was designed using a single-dose, 3-treatment cross-over trial in 50 subjects to predict the PK profile of the Xarelto® IR tablet in the fed state. Overall, the results obtained from the IVIVC model were found to be comparable with that from the PBPK model. The outcome from both the model pointed to the positive food effect on the in vivo profile of the Riva. The developed models thus can be effectively extended to establish bioequivalence for the marketed and novel complex formulations of Riva such as amorphous solid dispersions.

In the remote sensing domain, it is crucial to complete semantic segmentation on the raster images, e.g., river, building, forest, etc, on raster images. A deep convolutional encoder--decoder (DCED) network is the state-of-the-art semantic segmentation method for remotely sensed images. However, the accuracy is still limited, since the network is not designed for remotely sensed images and the training data in this domain is deficient. In this paper, we aim to propose a novel CNN for semantic segmentation particularly for remote sensing corpora with three main contributions. First, we propose applying a recent CNN called a global convolutional network (GCN), since it can capture different resolutions by extracting multi-scale features from different stages of the network. Additionally, we further enhance the network by improving its backbone using larger numbers of layers, which is suitable for medium resolution remotely sensed images. Second, "channel attention'' is presented in our network in order to select the most discriminative filters (features). Third, "domain-specific transfer learning'' is introduced to alleviate the scarcity issue by utilizing other remotely sensed corpora with different resolutions as pre-trained data. The experiment was then conducted on two given datasets: (i) medium resolution data collected from Landsat-8 satellite and (ii) very high resolution data called the ISPRS Vaihingen Challenge Dataset. The results show that our networks outperformed DCED in terms of $F1$ for 17.48% and 2.49% on medium and very high resolution corpora, respectively.

This cross-country four-year longitudinal study investigated the development of processing speed throughout primary school education. The analyses were conducted on data collected from 441 schoolchildren in Grades 1, 2, 3 and 4 (aged 6.42 to 11.85 years) in the Kyrgyzstan and Russia. Mixed effects growth modeling was applied to estimate average and individual growth trajectories for processing speed at two cross-country samples. Latent class growth modeling was conducted to identify different types of growth trajectories for processing speed and to compare the distribution of the types within the analyzed samples. According to the results, processing speed significantly increases across primary school years. The trajectory is characterized by nonlinear changes with more intensive growth between Grades 1 and 2, which later slows down until Grade 4. There were no significant cross-country differences in the initial score of processing speed or developmental changes in processing speed across primary school years. The development of processing speed is described by a model with three quadratic growth types but minimally differs. It is concluded that in both samples, the development of processing speed may be characterized by homogeneity, with the most intensive growth from Grade 1 to Grade 2 and subsequent linear improvement until Grade 4.

Electrical remodelling as a result of the homeodomain transcription factor 2 (Pitx2)‐dependent gene regulation was linked to atrial fibrillation (AF) and AF patients with single nucleotide polymorphisms at chromosome 4q25 responded favorably to Class I antiarrhythmic drugs (AADs). The possible reasons behind this remain elusive. The purpose of this study was to assess the efficacy of AADs disopyramide, quinidine, and propafenone on human atrial arrhythmias mediated by Pitx2-induced remodelling, from a single cell to the tissue level, using drug binding models with multi-channel pharmacology. Experimentally calibrated populations of human atrial action potential (AP) models in both sinus rhythm (SR) and Pitx2-induced AF conditions were constructed by using two distinct models to represent morphological subtypes of AP. Multi-channel pharmacological effects of disopyramide, quinidine, and propafenone on ionic currents were considered. Simulated results showed that Pitx2-induced remodelling increased maximum upstroke velocity (dVdtmax) and conduction velocity (CV), and decreased AP duration (APD) and wavelength (WL). At the concentrations tested in this study, these AADs decreased dVdtmax and CV and prolonged APD in the setting of Pitx2-induced AF. Our findings of alterations in WL indicated that quinidine and disopyramide may be more effective against Pitx2-induced AF than propafenone by prolonging WL.

This paper identifies a criterion for choosing the largest set of rejected hypotheses in high-dimensional data analysis where Multiple Hypothesis testing is used in exploratory research to identify significant associations among many variables. The method neither requires predetermined thresholds for level of significance, nor uses presumed thresholds for false discovery rate. The upper limit for number of rejected hypotheses is determined by finding maximum difference between expected true hypotheses and false hypotheses among all possible sets of rejected hypotheses. Methods of choosing a reasonable number of rejected hypotheses and application to non-parametric analysis of ordinal survey data are presented.

An effective process monitoring strategy is a requirement for meeting the challenges posed by increasingly complex products and manufacturing processes. To address these needs, this study investigates a comprehensive scheme based on classical machine learning methods, deep learning algorithms, and feature extraction and selection techniques. In a first step, a novel deep learning architecture based on convolutional neural networks (CNN) and gated recurrent units (GRU) is introduced to predict the local weld quality based on mid-wave infrared (MWIR) and near-infrared (NIR) image data. The developed technology is used to discover critical welding defects including lack of fusion (false friends), sagging and lack of penetration, and geometric deviations of the weld seam. Additional work is conducted to investigate the significance of various geometrical, statistical, and spatio-temporal features extracted from the keyhole and weld pool regions. Furthermore, the performance of the proposed deep learning architecture is compared to that of classical supervised machine learning algorithms, such as multi-layer perceptron (MLP), logistic regression (LogReg), support vector machines (SVM), decision trees (DT), random forest (RF) and k-Nearest Neighbors (kNN). Optimal hyperparameters for each algorithm are determined by an extensive grid search. Ultimately, the three best classification models are combined into an ensemble classifier that yields the highest detection rates and achieves the most robust estimation of welding defects among all classifiers studied, which is validated on previously unknown welding trials.

Jingmen viruses are newly described segmented flavi-like viruses that have a worldwide distribution in ticks and have been associated with febrile illnesses in humans. Computational analyses were used to predict that Jingmen flavi-like virus glycoproteins have structural features of class II viral fusion proteins, including an ectodomain consisting of beta-sheets and short alpha-helices, a fusion peptide with interfacial hydrophobicity and a three domain architecture. Jingmen flavi-like virus glycoproteins have a sequence enriched in serine, threonine and proline at the amino terminus, which is a feature of mucin-like domains. Several of the serines and threonines are predicted be modified by the addition of O-linked glycans. Some of the glycoproteins are predicted to have an additional mucin-like domain located prior to the transmembrane anchor, whereas others are predicted to have a stem consisting of two alpha-helices. The flavivirus envelope protein and Jingmen flavi-virus glycoproteins may have diverged from a common class II precursor glycoprotein with a mucin-like domain or domains acquired after divergence.

Both SARS-CoV-2 and SARS-CoV initially appeared in China and spread to other parts of the world. SARS-CoV-2 has generated a COVID-19 pandemic causing more than 6 million human deaths worldwide while the SARS outbreak quickly ended in six months with a global total of 774 reported deaths. One of the factors contributing to this stunning difference in the outcome between these two outbreaks is the inaccuracy of the RT-qPCR tests for SARS-CoV-2, which generated a large number of false-negative and false-positive test results that have misled patient management and public health policy-makers. This article presented Sanger sequencing evidence to show that the RT-PCR diagnostic protocol established in 2003 for SARS-CoV can in fact detect SARS-CoV-2 accurately due to the well-known nonspecific PCR amplification of DNAs with similar sequences. Using nested RT-PCR followed by Sanger sequencing to retest 50 patient samples collected in January, 2022 and sold as RT-qPCR positive reference confirmed 21 (42%) were false-positive. Although the other 29 positive isolates were categorized as Omicron variant by partial sequencing of the N gene, and the RBD and the NTD of the S gene, 9 (31%) showed focal to complete sequencing failure in the S gene segments due to multi-allelic SNPs. During the course of the study, an Omicron variant isolate containing a BA.1 NTD and a BA.2 RBD in its S gene was also detected. Routine partial S gene sequencing of all PCR-positive samples can timely discover multi-allelic SNPs and viral recombination in the circulating variants for investigation of their impacts on vaccine efficacies, therapeutics and diagnostics.

The primary screening of hybridoma cells is a time-critical and laborious step during the development of monoclonal antibodies. Often critical errors occur in this phase, which supports the notion that the generation of monoclonal antibodies with hybridoma technology is difficult to control and hence a risky venture. We think that it is crucial to improve the screening process to eliminate most of the immanent deficits of the conventional approach. With this new microarray-based procedure, several advances could be achieved: Selectivity for excellent binders, high throughput, reproducible signals, avoidance of misleading avidity (multivalency) effects, and simultaneous performance of competition experiments. The latter can directly be used to select clones of desired cross-reactivity properties. In this paper, a model system with two excellent clones against carbamazepine, two weak clones and blank supernatant has been designed to examine the effectiveness of the new system. The excellent clones could be detected largely independent of the IgG concentration, which is unknown during the clone screening since the determination and subsequent adjustment of the antibody concentration is not possible in most cases. Furthermore, in this approach, the enrichment, isolation, and purification of IgG for characterization is not necessary. Raw cell culture supernatant can be used directly, even when fetal calf serum (FCS) or other complex media had been used. In addition, an improved method for the oriented antibody-immobilization on epoxy-silanized slides is presented. Based on the results of this model system, we conclude that this approach should be preferable to most other protocols leading to many of false positives, causing expensive and lengthy confirmation steps to weed out the poor clones.

The prim objective of commenced article is to determine q-sumudu transforms of a product of unified family of q-polynomials with basic (or q-) analogue of fox’s H-function and q-analog of I-functions. Specialized cases of the leading outcome are further evaluated as q-sumudu transform of general class of q-polynomials and q-sumudu transforms of the basic analogues of Fox’s H-function and I-functions.

Exercise-based cardiac rehabilitation requires patients to perform a set of certain prescribed exercises a specific number of times. Local muscular endurance (LME) exercises are an important part of the rehabilitation program. Automatic exercise recognition and repetition counting, from wearable sensor data is an important technology to enable patients to perform exercises independently in remote settings, e.g. their own home. In this paper we first report on a comparison of traditional approaches to exercise recognition and repetition counting, corresponding to supervised machine learning and peak detection from inertial sensing signals respectively, with more recent machine learning approaches, specifically Convolutional Neural Networks (CNNs). We investigated two different types of CNN: one using the AlexNet architecture, the other using time-series array. We found that the performance of CNN based approaches were better than the traditional approaches. For exercise recognition task, we found that the AlexNet based single CNN model outperformed other methods with an overall 97.18% F1-score measure. For exercise repetition counting , again the AlexNet architecture based single CNN model outperformed other methods by correctly counting repetitions in 90% of the performed exercise sets within an error of ±1. To the best of our knowledge, our approach of using a single CNN method for both recognition and repetition counting is novel. In addition to reporting our findings, we also make the dataset we created, the INSIGHT-LME dataset, publicly available to encourage further research.

One important motivation for the implementation of the Water Framework Directive is the creation of non-market environmental benefits such as improved ecological quality, or greater opportunities for open-access river recreation via microbial pollution remediation. Pollution sources impacting on ecological or recreational water quality can be uncorrelated but non-market benefits arising from riverine improvements are typically conflated within benefit valuation studies. Using stated preference choice experiments, we seek to disaggregate these sources of value for different river users, thereby allowing decision makers to understand the consequences of adopting alternative investment strategies. Our results suggest anglers derive greater value from improvements to the ecological quality of river water, in contrast to swimmers and rowers for whom greater value is gained from improvements to recreational quality. We also find three distinct groups of respondents: a majority preferring ecological over recreational improvements, a substantial minority holding opposing preference orderings and a small proportion expressing relatively low values for either form of river quality enhancement. As such, this research demonstrates that the non-market benefits which may accrue from different types of water quality improvements are nuanced in terms of their potential beneficiaries and, by inference, their overall value and policy implications.

For esophageal cancer patients treated with neoadjuvant chemoradiotherapy (nCRT), restaging using F-18-fluorodeoxyglucose (18F-FDG) positron emission tomography computed tomography (PET-CT) following nCRT can detect interval metastases, including liver metastases, in almost 10% of patients. However, in clinical practice, focal FDG liver uptake, unrelated to liver metastases, is observed after chemoradiotherapy. This radiation-induced liver injury (RILI) can potentially lead to overstaging. Liver radiation damage is also a concern during distal esophageal cancer radiotherapy due to proximity to the left liver lobe, typically included in the radiation field. Post-CRT, if FDG activity appears in the left or caudate liver lobes, thorough investigation is needed to confirm or rule out distant metastases. The increased FDG uptake in liver lobes post-CRT often presents a diagnostic dilemma. Distinguishing between radiation-induced liver disease and metastasis is vital for appropriate patient management, necessitating a combination of imaging techniques and an understanding of the factors influencing radiation response. Diagnosis involves identifying new foci of hepatic FDG avidity on PET/CT scans. Geographic regions of hypoattenuation on CT and well-demarcated regions with specific enhancement patterns on contrast-enhanced CT scan and MRI are characteristic of radiation-induced liver disease (RILD). Lack of mass effect on all three modalities (CT, MRI, PET) indicates RILD. Resolution of abnormalities on subsequent examinations also helps in diagnosing RILD. Moreover, it can also help to rule out occult metastases, thereby excluding those patients from further surgery who will not benefit from esophagectomy with curative intent.

Short-range Internet of Things (IoT) sensor nodes operating at 2.4 GHz need to provide ubiquitous wireless sensor networks (WSNs) with energy-efficient, wide-range output power (POUT) and fully integrated on a single chip for wireless body area networks (WBANs) and wireless personal area networks (WPANs) using low-power Bluetooth (BLE) and Zigbee standards. The proposed fully integrated transmitter (TX) utilizes a digitally controllable current-mode class-D (CMCD) power amplifier (PA) with a second harmonic distortion (HD2) suppression to reduce VCO pulling in an integrated system while meeting harmonic limit regulations. The CMCD PA is divided into 7-bit slices that can be reconfigured between differential and single-ended topologies. Duty cycle distortion compensation is performed for HD2 suppression, and an HD2 rejection filter and a modified C-L-C low-pass filter (LPF) reduce HD2 further. Implemented in a 28 nm CMOS process, the TX achieves a wide POUT range of from 12.1 to -31 dBm and provides a maximum efficiency of 39.8% while consuming 41.1 mW at 12.1 dBm POUT. The calibrated HD2 level is -82.2 dBc at 9.93 dBm POUT, resulting in a transmitter figure of merit (TX_FoM) of -97.52 dB. Higher order harmonic levels remain below -41.2 dBm even at 12.1 dBm POUT, meeting regulatory requirements.

Quantum mechanics could be studied through polynomial algebra, as has been demonstrated by a work (“An approach to first-principles electronic structure calculation by symbolic-numeric computation” by A. Kikuchi). We carry forward the algebraic method of quantum mechanics through algebraic number theory; the basic equations are represented by the multivariate polynomial ideals; the symbolic computations process the ideal and disentangle the eigenstates as the algebraic variety; upon which one canbuild the Galois extension of the number field, in analogy with the univariate polynomial case, to investigate the hierarchy of solutions; the Galois extension is accompanied with the group operations, which permute the eigenstates from one to another, and furnish the quantum system with a non-apparent symmetry. Besides, this sort of algebraic quantum mechanics is an embodiment of the class field theory; some of the important consequences of the latter emerge in quantum mechanics. We shall demonstrate these points through simple models; we will see the use of computational algebra facilitates such sort of analysis, which might often be complicated if we try to solve them manually.

Continual learning, a crucial facet of machine learning, involves the perpetual acquisition of valuable insights from incoming data, sans the necessity for full dataset access. Esteemed as a fundamental goal in artificial intelligence, continual learning grapples with an enduring challenge—catastrophic forgetting. A proficient model must exhibit adaptability for new data assimilation and robustness to retain existing knowledge. Class-incremental learning (CIL) aids the gradual integration of knowledge from newly introduced classes, forming a universal classifier. However, directly training the model with fresh class instances triggers a problem—forgetting distinguishing features of prior classes, causing a performance decline. Addressing such issues in machine learning, this survey aims to delineate significant challenges, outcomes, and recent advancements, including our contributions to CIL techniques, especially in image classification.