Kil To Chong’s research while affiliated with Jeonbuk National University and other places

... The overall system framework is depicted in Figure D, integrates the simulated environment, image based scene analysis and emergency alert generation to enable realtime traffic incident response. This framework represents a significant advancement by building upon our previous work for developing a practical and efficient solution for highway accident management [65]. The simulation analysis section as illustrated in Figure D comprises of four main subsections: (1) LLM for Command Generation: The GPT 4-Turbo API processes emergency transcripts to generate precise commands for autonomous drone control commands including navigation and location based-specific operations. ...

Reference:

Transforming Highway Safety With Autonomous Drones and AI: A Framework for Incident Detection and Emergency Response

... The obtained binding scores were evaluated to identify the most favorable configurations, and the results were compared with available experimental data to validate the predictions. Additional simulations were considered to confirm the predicted interactions, and the results were interpreted in terms of potential mechanisms of action of Calothrixin A in the insect olfactory system [60][61][62]. ...

Reference:

Development and Perfection of Marine-Based Insecticide Biofilm for Pea Seed Protection: Experimental and Computational Approaches

... In addition, in deep convolutional networks, the back layer is mainly responsible for extracting deep abstract features, but to some ex-tent, it is insufficiently combined with the low-level features of the front layer, which leads to insufficient ability to capture certain detailed features [13]. For this reason, the introduction of an attention mechanism [14] is expected to enhance the network's ability to jointly capture global and local information, thus better supporting complex recognition tasks [15]. Furthermore, significant advancements have been made in large-scale character recognition in other countries. ...

Reference:

Combining Multi-Scale Fusion and Attentional Mechanisms for Assessing Writing Accuracy

... On the other hand, with the rapid advancements in artificial intelligence, there is an emergence in its utilization for image segmentation [15], [16], medical image analysis [17], [18], computational bio-informatics [19]- [21], object detection [22] and visual scene understanding using visual language models. Integrating the collaborative perception, YOLO models, Large Language Models (LLMs), and Vision-Language Models (VLMs) offer transformative potential for enhancing road safety [23], [24]. ...

Reference:

Transforming Highway Safety With Autonomous Drones and AI: A Framework for Incident Detection and Emergency Response

... Additionally, the typical object detection performance in agriculture ranges from 0.7 to 0.9 in F1-score (Rai et al., 2023;Ruigrok et al., 2023;Rehman et al., 2024), which is comparable with the detection performance observed in our experiments. This indicates that the orthomosaic provides an accurate simulation of images taken in a real-world flight. ...

Reference:

Adaptive path planning for efficient object search by UAVs in agricultural fields

... Additionally, the base estimators are paired with a final estimator, also known as Meta learner. As the name suggests final estimator it is responsible for taking the most accurate prediction from the base estimators as input and ultimately producing the final predictions [45][46][47]. While constructing the stacking classifier the mechanism involves the two stages of the training process. ...

Reference:

Analysis of Ruddlesden‐Popper and Dion‐Jacobson 2D Lead Halide Perovskites Through Integrated Experimental and Computational Analysis

... ProtGPT2 [21], a GPT2-based language model, has been trained on protein sequence space to generate novel proteins adhering to natural protein principles. Furthermore, fine-tuning pre-trained language models on specific datasets has been successfully applied in tasks such as enzyme design [22] and post-translational modification prediction [23], making them ideal for conotoxin generation. ...

Reference:

ConoGPT: Fine-Tuning a Protein Language Model by Incorporating Disulfide Bond Information for Conotoxin Sequence Generation

... Host phenotype classification mainly focuses on age prediction [24], cancer prediction [31], and disease condition prediction [27]. For drug response prediction, Maryam et al. utilized graph neural network (GNN) models to screen drugs inf luenced by the gut microbiome, contributing to the optimization of drug dosage and formulation [64]. Jiangsheng Pi et al. developed MDGNN, a deep learning algorithm based on graph convolutional networks (GCNs) to predict microbe-drug associations, successfully identifying two drugs associated with SARS-CoV-2 [61]. ...

Reference:

Techniques for learning and transferring knowledge for microbiome-based classification and prediction: review and assessment

... This study used seven different feature descriptors: AAC, PAAC, APAAC, SOCN, DPC, CKSAAGP, and GAAC, alongside five distinct classifiers: RF, XGB, LGBM, ETC, and SVM, for constructing prediction models. Machine learning, deep learning, and graph neural networks are extensively employed in the fields of bioinformatics Mir et al. 2023;Hassan et al. 2024b;Rehman et al. 2024, chemoinformatics Ahmad et al. 2023Mir et al. 2024, and image analysis Noor et al. 2020;Komura and Ishikawa 2018;Eesaar et al. 2023. We used the Optuna framework Akiba et al. 2019 to optimize each baseline model against every descriptor. ...

Reference:

Possum: identification and interpretation of potassium ion inhibitors using probabilistic feature vectors

... Among these, 7 articles [257,6,360,295,357,37,207] focus solely on antimicrobial peptides. Additionally, two review articles [124,300] have been published for cell-penetrating peptides, two for anti-viral peptides [17,188], and three for anti-cancer peptides [271,286,34]. For therapeutic [220], anti-inflammatory [261], and bitter peptides [218], a single review article has been published for each type. ...

Reference:

Peptide classification landscape: An in-depth systematic literature review on peptide types, databases, datasets, predictors architectures and performance