Using traditional methods based on detection rules written by human security experts, there is a significant challenge in accurately detecting many network threats that are increasingly becoming sophisticated. In order to deal with the limitation of the traditional methods, network threat detection techniques utilizing artificial intelligence technologies such as machine learning are being extensively studied. Research has also been conducted on analyzing various string patterns in network packet payloads through natural language processing techniques to detect attack intent. However, due to the nature of packet payloads containing binary data as well as text data, a new approach is needed that goes beyond typical natural language processing techniques. In this paper, we study a token extraction method optimized for payloads using n-gram and byte pair encoding techniques. Furthermore, we generate embedding vectors that can understand the context of the packet payload using algorithms such as Word2Vec and FastText. We also compute the embedding of various header data associated with packets, such as IP addresses and ports. Given the combination of these features, we ensemble a text 1D-CNN and a multi-head attention network in a novel fashion. We validated the effectiveness of our classification technique with the CICIDS2017 open dataset and over half a million data collected by The Education Cyber Security Center (ECSC) currently operating in South Korea. The proposed model showed remarkable progress compared to previous studies, demonstrating a highly accurate classification performance with an F1-Score of 0.998. Our model can also preprocess and classify 150,000 network threats per minute to help security agents in the field divert their time to analyzing more complex attack patterns.

Interior permanent magnets motors are widely used in applications requiring high power den-sity and high efficiency due to their high torque generating capabilities. Recently, given the re-cent price fluctuations and unstable supply of rare-earth permanent magnets, alternative config-urations with reduced use of permanent magnets are being sought. Among the various candi-dates related to this, the consequent-pole type rotor structure can halve the number of perma-nent magnets used compared with conventional structures. However, in the no-load analysis, the waveform of the back electromotive force becomes asymmetric, generating a harmonic component. As a result, there is a disadvantage that the torque ripple increases. To overcome these shortcomings, we propose a novel rotor structure that applies a consequent-pole structure to an embedded permanent-magnet motor structure, wherein a number of permanent magnets are arranged in a flared structure to constitute a single polarity. In the proposed flared-structured magnet arrangement, it is possible to adjust the angle of the permanent magnet and the polar angle to mitigate the asymmetry of the back EMF waveform. The proposed struc-ture was optimized with a genetic algorithm and a prototype of the optimal model was con-structed and experimentally evaluated to verify its validity. Finally, the performance improve-ment and validity of the proposed structure were verified by comparing the analysis results of the optimal model with the experimental results.

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.

Background/Objectives: This study aims to investigate the influence of root canal morphology and various treatment variables on the outcomes of root canal treatments (RCT) in mandibular second molars, assessed through cone-beam computed tomography (CBCT) imaging. Methods: A total of 150 CBCT images were examined, comprising 100 cases of persistent endodontic infections and 50 from previously treated root canals with normal apices in mandibular second molars. CBCT was utilized to evaluate root canal configuration, the radiographic quality of coronal restorations and treated canal systems, and the presence of periapical lesions. Statistical analyses were performed to explore correlations between these factors. Results: The presence of a C-shaped root canal configuration did not demonstrate a significant correlation with periapical lesions (p=0.05). Factors influencing endodontic treatment outcomes included the presence of missing canals (p=0.018), underfilling or overfilling (p=0.045), and inadequate coronal restoration (p=0.006). The presence of a missing canal was identified as the variable most significantly associated with periapical lesions (OR=3.103). Leaky canals were more commonly observed in C-shaped root canals (p<0.001). Conclusions: Regardless of the root canal morphology of mandibular second molars, successful RCT depends on thorough disinfection to eliminate any untreated canals, precise three-dimensional filling of the canals at the correct working length, and a securely sealed coronal restoration to prevent leakage.

White spot lesions (WSLs), a side effect of orthodontic treatment, can result in reversible and unaesthetic results. Graphene oxide (GO) with a bioactive glass (BAG) mixture(BAG@GO) was added to Low Viscosity Transbond XT(LV) in a ratio of 1, 3, 5%. The composite’s characterization and its physical and biological properties were verified with scanning electron microscopy(SEM) and X-ray diffraction(XRD); its microhardness, shear bond stress (SBS), cell viability, and adhesive remnant index (ARI) were also assessed. Efficiency in reducing WSL was evaluated using antibacterial activity of S. mutans. Anti-demineralization was analyzed using a cycle of the acid-base solution. Adhesives with 3 or 5 wt.% of BAG@GO showed significant increase in microhardness compared with LV. The sample and LV groups showed no significant differences in SBS or ARI. The cell viability test confirmed that none of the sample groups showed higher toxicity compared to the LV group. Antibacterial activity was higher in the 48-hour group than in the 24-hour group; the 48-hour test showed that BAG@GO had a high antibacterial effect, which was more pronounced in 5 wt.% of BAG@GO. Anti-demineralization effect was higher in the BAG@GO-group than in the LV-group; the higher the BAG@GO concentration, the higher the anti-demineralization effect.

Definitive surgical resection is the preferred treatment for early-stage non-small-cell lung cancer (NSCLC). Research into genetic alterations, including epidermal growth factor receptor (EGFR) mutation, in early stage NSCLC remains insufficient. Here, we investigated the prevalence of genetic alterations in early-stage NSCLC and the association between EGFR mutation and recurrence after complete resection. Between January 2019 and December 2021, 659 patients with NSCLC who underwent curative surgical resection at a single regional cancer center were recruited. We compared the clinical and pathological data between the recurrence and non-recurrence groups. Multivariate logistic regression was used to predict the risk factors for recurrence. Among the 659 enrolled cases, the most common histology was adenocarcinoma (74.5%), followed by squamous cell carcinoma (21.7%). The prevalence of EGFR mutation was 43% (194/451). Among them, L858R point mutation and exon 19 deletion was 52.3% and 42%, respectively. ALK rearrangement was found at 5.7% (26/453), and ROS1 fusion was found at 1.6% (7/441). The recurrence rate of the entire population was 19.7%. In multivariate analysis, the presence of EGFR mutation, stage II or III (vs. stage I), and pathologic subtype (presence of solid type) were associated with recurrence. Among the recurred group, 86.5% of the patients with EGFR mutation experienced distant recurrence compared to only 66.7% of wild-type (p = 0.016), with no significant difference in median disease-free survival (p = 0.983). In conclusion, the prevalence of EGFR mutation, ALK rearrangement, and ROS1 fusion was 43.0%, 5.7%, and 1.6%, respectively in patients with early-stage NSCLC who underwent curative resection. Along with stage II/III and solid pathologic subtype, EGFR mutation was an independent risk factor for recurrence. In the recurrence group, the rate of distant metastasis was higher in patients with EGFR mutation than in those with wild-type.