Intelligent Systems with Applications最新文献_第4页

Emulating fundamental analysts: Analytical stage-based multi-agent framework enhanced with expert guidance and Preference-Anchored Likelihood Adjustment 模拟基本面分析师：基于分析阶段的多代理框架，通过专家指导和偏好附加可能性调整得到增强

Intelligent Systems with Applications Pub Date : 2025-02-27 DOI: 10.1016/j.iswa.2025.200496

Tao Xu , Zhe Piao , Tadashi Mukai , Yuri Murayama , Kiyoshi Izumi

{"title":"Emulating fundamental analysts: Analytical stage-based multi-agent framework enhanced with expert guidance and Preference-Anchored Likelihood Adjustment","authors":"Tao Xu , Zhe Piao , Tadashi Mukai , Yuri Murayama , Kiyoshi Izumi","doi":"10.1016/j.iswa.2025.200496","DOIUrl":"10.1016/j.iswa.2025.200496","url":null,"abstract":"<div><div>With the rapid advancement of large language models (LLMs), some studies have explored their potential for predicting stock prices based on financial texts. However, previous research often overlooked the depth of analysis generated by LLMs, resulting in reasoning processes inferior to those of human analysts. In fundamental investing, which requires in-depth company analysis, conclusions from imperfect reasoning lack persuasiveness. In this study, inspired by the analysis process of human analysts, we propose an “Analytical Stage-Based Multi-Agent Framework” to enable LLMs to perform in-depth fundamental analysis. This framework divides the analysis into multiple stages, assigning an LLM agent to each. We enhance each agent’s capabilities for its specific task through expert guidance or fine-tuning, allowing them to collectively emulate the workflow of human analysts. Furthermore, we introduce Preference-Anchored Likelihood Adjustment, a new method for fine-tuning LLMs. This approach addresses the decline in likelihood of generating correct responses that occurs after using existing preference alignment methods. It employs an objective function with two terms: one to increase likelihood and another to preserve aligned preference. We conducted experiments using our framework to analyze company earnings releases. We evaluated the analysis quality based on comprehensiveness and logical soundness, while correctness was assessed by using stock prices as the ground truth to calculate the Matthews correlation coefficient and F1 score. Results demonstrate that even without expert guidance and fine-tuning, our multi-agent framework can enhance LLMs in both analysis quality and correctness. When combined with expert guidance and fine-tuning, the performance is further improved.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200496"},"PeriodicalIF":0.0,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143529747","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-Agent Reinforcement Learning for Cybersecurity: Classification and survey 面向网络安全的多智能体强化学习：分类与综述

Intelligent Systems with Applications Pub Date : 2025-02-25 DOI: 10.1016/j.iswa.2025.200495

Salvo Finistrella, Stefano Mariani, Franco Zambonelli

{"title":"Multi-Agent Reinforcement Learning for Cybersecurity: Classification and survey","authors":"Salvo Finistrella, Stefano Mariani, Franco Zambonelli","doi":"10.1016/j.iswa.2025.200495","DOIUrl":"10.1016/j.iswa.2025.200495","url":null,"abstract":"<div><div>In the face of a rapidly evolving threat landscape, traditional cybersecurity measures – such as signature-based detection and static rules on firewalls, intrusion detection systems (IDS) and antivirus software – often lag behind sophisticated cyber attacks. Through a review of existing literature, we examine the shortcomings of traditional cybersecurity methods and how these can be surpassed with the application of Reinforcement Learning (RL) based methods. This study classifies RL-based approaches to cybersecurity, aimed at enhancing detection, mitigation and response to cyber attacks, along two orthogonal dimensions: the RL Frameworks used (e.g. single-agent vs. multi-agent) and the network configuration where they are deployed (e.g. host-based, or network-based cybersecurity). The goal is that of aiding researchers and practitioners interested in the field to quickly understand what are the opportunities for RL-based cybersecurity depending on the network environment to be protected and point them to the representative articles in the field. Finally, we emphasize the importance of further research and development to address challenges such as computational complexity, generalization and data quality.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200495"},"PeriodicalIF":0.0,"publicationDate":"2025-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143521241","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

EHR-protect: A steganographic framework based on data-transformation to protect electronic health records EHR-protect：基于数据转换的隐写框架，用于保护电子健康记录

Intelligent Systems with Applications Pub Date : 2025-02-24 DOI: 10.1016/j.iswa.2025.200493

Adifa Widyadhani Chanda D'Layla , Ntivuguruzwa Jean De La Croix , Tohari Ahmad , Fengling Han

{"title":"EHR-protect: A steganographic framework based on data-transformation to protect electronic health records","authors":"Adifa Widyadhani Chanda D'Layla , Ntivuguruzwa Jean De La Croix , Tohari Ahmad , Fengling Han","doi":"10.1016/j.iswa.2025.200493","DOIUrl":"10.1016/j.iswa.2025.200493","url":null,"abstract":"<div><div>The increasing digitization of healthcare systems and the shift to Electronic Health Records (EHRs) have introduced critical security challenges, including unauthorized access, data breaches, and confidentiality risks. For example, the rapid exchange of sensitive health data between stakeholders highlights the need for secure data-sharing mechanisms. To address these challenges, steganography emerges as a critical solution by embedding sensitive information within other data forms, reducing the likelihood of unauthorized access and ensuring patient confidentiality. This study presents EHR-Protect, an innovative steganographic framework designed to secure EHRs by embedding them within medical images. Unlike general-purpose images, medical images are susceptible to distortions as they serve as diagnostic tools. EHR-Protect uses logarithmic pixel transformation and adaptive techniques such as difference expansion and EHR magnitude reduction to minimize distortions in carrier medical images. The results of EHR-Protect demonstrate its effectiveness in securely embedding EHRs into medical images with minimal distortions. The proposed method achieves a high Peak Signal-to-Noise Ratio (PSNR) of 91.90 dB and a perfect Structural Similarity Index Measure (SSIM) of 1, ensuring image quality is maintained. MSE values across different cover images show minimal increases, even as secret data payloads rise from 10 to 100 kilobits, indicating controlled distortion. The results confirm that EHR-Protect outperforms existing methods, offering a robust solution for securing the EHR data without compromising medical image integrity.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200493"},"PeriodicalIF":0.0,"publicationDate":"2025-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143549361","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Binary classification with Fuzzy-Bayesian logistic regression using Gaussian fuzzy numbers 基于高斯模糊数的模糊贝叶斯逻辑回归二元分类

Intelligent Systems with Applications Pub Date : 2025-02-24 DOI: 10.1016/j.iswa.2025.200494

Georgios Charizanos , Haydar Demirhan , Duygu İçen

{"title":"Binary classification with Fuzzy-Bayesian logistic regression using Gaussian fuzzy numbers","authors":"Georgios Charizanos , Haydar Demirhan , Duygu İçen","doi":"10.1016/j.iswa.2025.200494","DOIUrl":"10.1016/j.iswa.2025.200494","url":null,"abstract":"<div><div>Binary classification is a critical task in pattern recognition applications in artificial intelligence and machine learning. The main weakness of binary classifiers is their sensitivity towards the imbalance in the number of observations in the binary classes and separation by a subset of features. Although various robust approaches are introduced against these issues, they need prolonged runtimes, limiting their applicability in artificial intelligence applications or for large datasets. In this study, we introduce a new binary classification framework called the fuzzy-Bayesian logistic regression, which incorporates robust Bayesian logistic regression with fuzzy classification using Gaussian fuzzy numbers. The proposed method improves classification performance while providing significant gains in computation time. We benchmark the proposed method with eight fuzzy, Bayesian, and machine learning classifiers using seventeen datasets. The results indicate that the fuzzy-Bayesian logistic regression outperforms all benchmark methods across all datasets in terms of six performance indicators. Moreover, the proposed method is shown to be significantly more efficient than its closest competitor, improving computational efficiency. The proposed method provides a promising binary classifier for a wide range of applications with its computational efficiency and robustness towards imbalance and separation issues in the data.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200494"},"PeriodicalIF":0.0,"publicationDate":"2025-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143512461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A multi-class hybrid variational autoencoder and vision transformer model for enhanced plant disease identification 一种多类混合变分自编码器和视觉变压器模型，用于增强植物病害识别

Intelligent Systems with Applications Pub Date : 2025-02-19 DOI: 10.1016/j.iswa.2025.200490

Folasade Olubusola Isinkaye , Michael Olusoji Olusanya , Ayobami Andronicus Akinyelu

{"title":"A multi-class hybrid variational autoencoder and vision transformer model for enhanced plant disease identification","authors":"Folasade Olubusola Isinkaye , Michael Olusoji Olusanya , Ayobami Andronicus Akinyelu","doi":"10.1016/j.iswa.2025.200490","DOIUrl":"10.1016/j.iswa.2025.200490","url":null,"abstract":"<div><div>Agriculture is considered as the propeller of economic growth as it accounts for 6.4 % of global gross domestic product (GDP) and in low-income countries, it can account for more than 25 % of GDP. Plants supply more than 80 % of the food consumed by humans and are the main source of nutrition for animals. Plant diseases pose a major risk to global food security as they account for losses of between 10 to 30 % of the global harvest every year. Deep learning techniques like convolutional neural networks successfully identify image-based diseases but struggle with capturing long-range contextual information. This makes them less robust in noisy or high-resolution images. Their high computational and memory demands also limit scalability for large datasets. To overcome these issues, we propose a hybrid model with the potential to combine Variational Autoencoders and Vision Transformers for enhanced accuracy and robustness of plant disease classification. Variational Autoencoder reduces image dimensionality while preserving essential features, and Vision Transformer captures global relationships to enhance accuracy and scalability, especially in multi-class disease classification. The experiment used images of corn, potato, and tomato plant leaves from the publicly available PlantVillage dataset. On-the-fly data augmentation was applied to further increase the robustness of the model. The proposed model achieved a classification accuracy of 93.2 %. This technique provides a reliable and efficient solution for identifying multiple plant diseases across various crops. It enhances agricultural productivity and supports food security efforts.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200490"},"PeriodicalIF":0.0,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143549363","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Content moderation assistance through image caption generation 通过图像标题生成帮助内容审核

Intelligent Systems with Applications Pub Date : 2025-02-16 DOI: 10.1016/j.iswa.2025.200489

Liam Kearns

{"title":"Content moderation assistance through image caption generation","authors":"Liam Kearns","doi":"10.1016/j.iswa.2025.200489","DOIUrl":"10.1016/j.iswa.2025.200489","url":null,"abstract":"<div><div>The rapid growth in digital media creation has led to an increased challenge in content moderation. Manual and automated moderation are susceptible to risks associated with a slower response time and false positives arising from unpredictable user inputs respectively. Image caption generation has been suggested as a viable content moderation tool, but there is a lack of real world deployment in this context. In this work, a collaborative approach is taken, where a machine learning model is used to assist human moderators in the approval and rejection of media within a scavenger hunt game. The proposed model is trained on the Flickr30k and MS Coco datasets to generate captions for images. The results demonstrate a 13% reduction in review times, indicating that human–machine collaboration contributes to mitigating the risk of unsustainable review backlog growth. Furthermore, fine-tuning the model led to a 28% reduction in review times when compared to the untuned model. Notably, this paper contributes to knowledge by demonstrating caption generation as a viable content moderation tool in addition to its sensitivity to accurate captions, whereby false positives risk a deterioration in moderator response time.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"25 ","pages":"Article 200489"},"PeriodicalIF":0.0,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143436567","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Innovative hybrid algorithm for efficient routing of limited capacity vehicles 有限容量车辆高效路径的创新混合算法

Intelligent Systems with Applications Pub Date : 2025-02-08 DOI: 10.1016/j.iswa.2025.200491

Vu Hong Son Pham , Van Nam Nguyen , Nghiep Trinh Nguyen Dang

{"title":"Innovative hybrid algorithm for efficient routing of limited capacity vehicles","authors":"Vu Hong Son Pham , Van Nam Nguyen , Nghiep Trinh Nguyen Dang","doi":"10.1016/j.iswa.2025.200491","DOIUrl":"10.1016/j.iswa.2025.200491","url":null,"abstract":"<div><div>This study addresses the critical challenges posed by the capacitated vehicle routing problem (CVRP), particularly in the logistics of cement transportation under capacity constraints. Existing algorithms, including grey wolf optimizer (GWO) and whale optimization algorithm (WOA), exhibit significant limitations such as imbalanced exploration and exploitation, inefficiency in refining solutions, and inadequate adaptability to dynamic routing conditions. These limitations hinder their ability to provide comprehensive solutions that optimize time, cost, and environmental sustainability. To address these critical challenges, this research proposes an enhanced hybrid metaheuristic algorithm, mGWOA, designed to overcome the limitations of existing approaches by combining the GWO's strong exploitation capabilities and the WOA's exploratory strengths. By integrating opposition-based learning (OBL) to expand the search space and mutation techniques to escape local optima, the mGWOA is tailored to provide more flexible, adaptive, and efficient solutions for the complex and dynamic requirements of the CVRP. The mGWOA framework leverages the exploratory advantages of WOA, the exploitative strengths of GWO, and the diversity-promoting features of OBL and mutation to address the complexities of CVRP. Through computational evaluations in various scenarios, including five case studies ranging from small to large, the algorithm demonstrates its superior ability to generate high-quality solutions, especially as the customer base expands. The results underscore the potential of mGWOA as a robust and adaptive approach to solving CVRP, minimizing time and cost, and contributing to sustainable logistics operations. By bridging existing knowledge gaps, this research provides an innovative global optimization framework, offering practical applications for CVRP and other engineering challenges.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"25 ","pages":"Article 200491"},"PeriodicalIF":0.0,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143421817","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Decoding legal processes: AI-driven system to streamline processing of the criminal records in Moroccan courts 解码法律程序：人工智能驱动的系统简化了摩洛哥法院对犯罪记录的处理

Intelligent Systems with Applications Pub Date : 2025-02-06 DOI: 10.1016/j.iswa.2025.200487

Taoufiq El Moussaoui, Chakir Loqman, Jaouad Boumhidi

{"title":"Decoding legal processes: AI-driven system to streamline processing of the criminal records in Moroccan courts","authors":"Taoufiq El Moussaoui, Chakir Loqman, Jaouad Boumhidi","doi":"10.1016/j.iswa.2025.200487","DOIUrl":"10.1016/j.iswa.2025.200487","url":null,"abstract":"<div><div>In Morocco, the manual process of feeding the criminal records database has become more challenging as the number of judgments has increased. This operation is carried out in two stages. The court clerk classifies the judgments as convictions or non-convictions, then extracts the guilty personal details and case information from those that present a conviction to feed the criminal records database. The current process has several drawbacks such as prolonged processing times, potential errors, and data confidentiality concerns. In this paper, we present a novel Arabic decision support legal system designed to assist in feeding the criminal records database. The system comprises two key components. The first component is a CNN-based judgment classifier that classifies judgments into convictions and non-convictions, while the second component is a legal entities extractor that can efficiently extract 11 entities from judgments classified as conviction. Both models were trained on purpose-built Arabic legal corpora created based on 4966 Arabic verdicts issued from the Moroccan courts. The judgment classifier achieves an accuracy of 96.6% on the judicial decision corpus, 98% on the Khaleej dataset, and 96.27% on the ECHR dataset. The legal entities extractor achieves 98.42%, 93.72%, and 93.5% F-scores on the legal entities corpus, the ANERCorp dataset, and the CONLL2003 respectively, outperforming prior research. These results highlight the potential of the system in improving the operation of feeding the criminal records database. Furthermore, the creation of these Arabic legal corpora provides valuable resources for enhancing legal document classification and domain-specific NER models in Arabic.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"25 ","pages":"Article 200487"},"PeriodicalIF":0.0,"publicationDate":"2025-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143396056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery LEAF-YOLO：轻型边缘实时航空图像小目标检测

Intelligent Systems with Applications Pub Date : 2025-02-01 DOI: 10.1016/j.iswa.2025.200484

Van Quang Nghiem, Huy Hoang Nguyen, Minh Son Hoang

{"title":"LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery","authors":"Van Quang Nghiem, Huy Hoang Nguyen, Minh Son Hoang","doi":"10.1016/j.iswa.2025.200484","DOIUrl":"10.1016/j.iswa.2025.200484","url":null,"abstract":"<div><div>Advances in Unmanned Aerial Vehicles (UAVs) and deep learning have spotlighted the challenges of detecting small objects in UAV imagery, where limited computational resources complicate deployment on edge devices. While many high-accuracy deep learning solutions have been developed, their large parameter sizes hinder deployment on edge devices where low latency and efficient resource use are essential. To address this, we propose LEAF-YOLO, a lightweight and efficient object detection algorithm with two versions: LEAF-YOLO (standard) and LEAF-YOLO-N (nano). Using Lightweight-Efficient Aggregating Fusion along with other blocks and techniques, LEAF-YOLO enhances multiscale feature extraction while reducing complexity, targeting small object detection in dense and varied backgrounds. Experimental results show that both LEAF-YOLO and LEAF-YOLO-N outperform models with fewer than 20 million parameters in accuracy and efficiency on the Visdrone2019-DET-val dataset, running in real-time (<span><math><mo>></mo></math></span>30 FPS) on the Jetson AGX Xavier. LEAF-YOLO-N achieves 21.9% AP<span><math><msub><mrow></mrow><mrow><mo>.</mo><mn>50</mn><mo>:</mo><mo>.</mo><mn>95</mn></mrow></msub></math></span> and 39.7% AP<span><math><msub><mrow></mrow><mrow><mo>.</mo><mn>50</mn></mrow></msub></math></span> with only 1.2M parameters. LEAF-YOLO achieves 28.2% AP<span><math><msub><mrow></mrow><mrow><mo>.</mo><mn>50</mn><mo>:</mo><mo>.</mo><mn>95</mn></mrow></msub></math></span> and 48.3% AP<span><math><msub><mrow></mrow><mrow><mo>.</mo><mn>50</mn></mrow></msub></math></span> with 4.28M parameters. Furthermore, LEAF-YOLO attains 23% AP<span><math><msub><mrow></mrow><mrow><mo>.</mo><mn>50</mn></mrow></msub></math></span> on the TinyPerson dataset, outperforming models with <span><math><mo>≥</mo></math></span> 20 million parameters, making it suitable for UAV-based human detection.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"25 ","pages":"Article 200484"},"PeriodicalIF":0.0,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Using Generative Adversarial Networks for the synthesis of emotional facial expressions in virtual educational environments 在虚拟教育环境中使用生成对抗网络合成情绪面部表情

Intelligent Systems with Applications Pub Date : 2025-02-01 DOI: 10.1016/j.iswa.2025.200479

William Villegas-Ch , Alexandra Maldonado Navarro , Araceli Mera-Navarrete

{"title":"Using Generative Adversarial Networks for the synthesis of emotional facial expressions in virtual educational environments","authors":"William Villegas-Ch , Alexandra Maldonado Navarro , Araceli Mera-Navarrete","doi":"10.1016/j.iswa.2025.200479","DOIUrl":"10.1016/j.iswa.2025.200479","url":null,"abstract":"<div><div>The generation of emotional facial expressions using Generative Adversarial Networks (GANs) has been widely researched, achieving significant advances in creating high-quality images. However, one of the main challenges remains the accurate transmission of complex and negative emotions, such as anger or sadness, due to the difficulty of correctly capturing the facial micro gestures that characterize these emotions. Traditional GAN architectures, such as StyleGAN and DCGAN, have proven highly effective in synthesizing positive emotions such as joy. However, they have limitations regarding more subtle emotions, leading to a higher rate of false negatives. In response to this problem, we propose fine-tuning the GAN discriminator. This tuning optimizes the discriminator’s ability to identify the minor details in facial expressions by using perceptual losses, allowing for better differentiation between the generated emotions, reducing classification errors, and improving precision in difficult-to-represent emotions. The results obtained in this study demonstrate a significant improvement in the system’s precision, especially in complex emotions. Precision for the anger emotion increased from 85.7% to 89.1%, and the number of false negatives was reduced from 16 to 10. Overall, precision for complex emotions exceeded 85%, substantially improving traditional solutions. These results demonstrate the potential of fine-tuning the GAN architecture for applications requiring more faithful and effective emotional interaction, significantly improving user experience across multiple domains.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"25 ","pages":"Article 200479"},"PeriodicalIF":0.0,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143303226","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0