PeerJ Computer Science最新文献

{"title":"Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach.","authors":"Shu Xu, Tao Wang, Zenghui Ding, Yu Wang, Tongsheng Wan, Dezhang Xu, Xianjun Yang, Ting Sun, Meng Li","doi":"10.7717/peerj-cs.2888","DOIUrl":"10.7717/peerj-cs.2888","url":null,"abstract":"<p><p>Biomechanical analysis of the human lower limbs plays a critical role in movement assessment, injury prevention, and rehabilitation guidance. Traditional gait analysis techniques, such as optical motion capture systems and biomechanical force platforms, are limited by high costs, operational complexity, and restricted applicability. In view of this, this study proposes a cost-effective and user-friendly approach that integrates inertial measurement units (IMUs) with a novel deep learning framework for real-time lower limb joint torque estimation. The proposed method combines time-frequency domain analysis through continuous wavelet transform (CWT) with a hybrid architecture comprising multi-head self-attention (MHSA), bidirectional long short-term memory (Bi-LSTM), and a one-dimensional convolutional residual network (1D Conv ResNet). This integration enhances feature extraction, noise suppression, and temporal dependency modeling, particularly for non-stationary and nonlinear signals in dynamic environments. Experimental validation on public datasets demonstrates high accuracy, with a root mean square error (RMSE) of 0.16 N·m/kg, Coefficient of Determination (<i>R</i> ²) of 0.91, and Pearson correlation coefficient of 0.95. Furthermore, the framework outperforms existing models in computational efficiency and real-time applicability, achieving a single-cycle inference time of 152.6 ms, suitable for portable biomechanical monitoring systems.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2888"},"PeriodicalIF":3.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192784/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499278","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BlockDroid: detection of Android malware from images using lightweight convolutional neural network models with ensemble learning and blockchain for mobile devices.","authors":"Emre Şafak, İbrahim Alper Doğru, Necaattin Barışçı, İsmail Atacak","doi":"10.7717/peerj-cs.2918","DOIUrl":"10.7717/peerj-cs.2918","url":null,"abstract":"<p><p>Due to the increase in the volume and diversity of malware targeting Android systems, research on detecting this harmful software is steadily growing. Traditional malware detection studies require significant human intervention and resource consumption to analyze all malware files. Moreover, malware developers have developed polymorphism and code obfuscation techniques to evade traditional signature-based detection approaches used by antivirus companies. Consequently, traditional methods have become increasingly inadequate for malware detection. So far, many machine learning methods have been successfully applied to address the issue of malware detection. Recent efforts in this area have turned to deep learning methods. Because these methods can automatically extract meaningful features from data and efficiently learn complex relationships, they can achieve better performance in malware detection as well as in solving many other problems. This article presents BlockDroid, an approach that combines convolutional neural network (CNN) models, ensemble learning, and blockchain technology to increase the accuracy and efficiency of malware detection for mobile devices. By converting Android DEX files into image data, BlockDroid leverages the superior image analysis capabilities of CNN models to discern patterns indicative of malware. The CICMalDroid 2020 dataset, comprising 13,077 applications, was utilized to create a balanced dataset of 3,590 images, with an equal number of benign and malware instances. The proposed detection system was developed using lightweight models, including EfficientNetB0, MobileNetV2, and a custom model as CNN models. Experimental studies were conducted by applying both individual models and the proposed BlockDroid system to our dataset. The empirical results illustrate that BlockDroid surpasses the performance of the individual models, demonstrating a substantial accuracy rate of 97.38%. Uniquely, BlockDroid integrates blockchain technology to record the predictions made by the malware detection model, thereby eliminating the need for re-analysis of previously evaluated applications and ensuring more efficient resource utilization. Our approach offers a promising and innovative strategy for effective and efficient Android malware detection.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2918"},"PeriodicalIF":3.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192715/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499219","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automatic detection of teacher behavior in classroom videos using AlphaPose and Faster R-CNN algorithms.","authors":"Jing Huang, Harwati Hashim, Helmi Norman, Mohammad Hafiz Zaini, Xiaojun Zhang","doi":"10.7717/peerj-cs.2933","DOIUrl":"10.7717/peerj-cs.2933","url":null,"abstract":"<p><p>This study proposes an automated classification framework for evaluating teacher behavior in classroom settings by integrating AlphaPose and Faster region-based convolutional neural networks (R-CNN) algorithms. The method begins by applying AlphaPose to classroom video footage to extract detailed skeletal pose information of both teachers and students across individual frames. These pose-based features are subsequently processed by a Faster R-CNN model, which classifies teacher behavior into appropriate or inappropriate categories. The approach is validated on the Classroom Behavior (PCB) dataset, comprising 74 video clips and 51,800 annotated frames. Experimental results indicate that the proposed system achieves an accuracy of 74.89% in identifying inappropriate behaviors while also reducing manual behavior logging time by 47% and contributing to a 63% decrease in such behaviors. The findings highlight the potential of computer vision techniques for scalable, objective, and real-time classroom behavior analysis, offering a viable tool for enhancing educational quality and teacher performance monitoring.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2933"},"PeriodicalIF":3.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12193412/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The application of blockchain technology in data trading: a systematic review.","authors":"Wei Xiong, Huaibin Shao, Hong Ge","doi":"10.7717/peerj-cs.2925","DOIUrl":"10.7717/peerj-cs.2925","url":null,"abstract":"<p><p>In the era of exponential data growth, the imperative to establish secure and efficient data trading mechanisms has become paramount. While traditional centralized architectures present critical limitations in security and operational efficiency, the emergence of blockchain technology offers transformative potential for decentralized solutions. This study conducts a systematic literature review to critically examine blockchain's evolving role in data trading ecosystems. Adhering to the PRISMA 2020 framework, we analyzed 18 rigorously selected studies from an initial pool of 164 Web of Science publications identified through \"data trading\" and \"blockchain\" keyword searches. Our analysis reveals three principal findings: first, current blockchain implementations predominantly cluster within computer science applications, indicating disciplinary concentration. Second, technical development emphasizes solution-oriented systems over theoretical model construction, suggesting an application-prioritized research paradigm. Third, we identify persistent challenges across three critical dimensions: (i) security-efficiency paradox in decentralized architectures, (ii) transparency-privacy equilibrium maintenance, and (iii) scalability constraints under high-concurrency scenarios. This study aims to offer in-depth insights into blockchain's potential applications in data trading and future research directions.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2925"},"PeriodicalIF":3.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192792/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499370","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An improved hippopotamus optimization algorithm based on adaptive development and solution diversity enhancement.","authors":"Shengyu Pei, Gang Sun, Lang Tong","doi":"10.7717/peerj-cs.2901","DOIUrl":"10.7717/peerj-cs.2901","url":null,"abstract":"<p><p>This study proposes an improved hippopotamus optimization algorithm to address the limitations of the traditional hippopotamus optimization algorithm in terms of convergence performance and solution diversity in complex high-dimensional problems. Inspired by the natural behavior of hippopotamuses, this article introduces chaotic map initialization, an adaptive exploitation mechanism, and a solution diversity enhancement strategy based on the original algorithm. The chaotic map is employed to optimize the initial population distribution, thereby enhancing the global search capability. The adaptive exploitation mechanism dynamically adjusts the weights between the exploration and exploitation phases to balance global and local searches. The solution diversity enhancement is achieved through the introduction of nonlinear perturbations, which help the algorithm avoid being trapped in local optima. The proposed algorithm is validated on several standard benchmark functions (CEC17, CEC22), and the results demonstrate that the improved algorithm significantly outperforms the original hippopotamus optimization algorithm and other mainstream optimization algorithms in terms of convergence speed, solution accuracy, and global search ability. Moreover, statistical analysis further confirms the superiority of the improved algorithm in balancing exploration and exploitation, particularly when dealing with high-dimensional multimodal functions. This study provides new insights and enhancement strategies for the application of the hippopotamus optimization algorithm in solving complex optimization problems.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2901"},"PeriodicalIF":3.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192830/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499191","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Proto-Caps: interpretable medical image classification using prototype learning and privileged information.","authors":"Luisa Gallée, Catharina Silvia Lisson, Timo Ropinski, Meinrad Beer, Michael Götz","doi":"10.7717/peerj-cs.2908","DOIUrl":"10.7717/peerj-cs.2908","url":null,"abstract":"<p><p>Explainable artificial intelligence (xAI) is becoming increasingly important as the need for understanding the model's reasoning grows when applying them in high-risk areas. This is especially crucial in the field of medicine, where decision support systems are utilised to make diagnoses or to determine appropriate therapies. Here it is essential to provide intuitive and comprehensive explanations to evaluate the system's correctness. To meet this need, we have developed Proto-Caps, an intrinsically explainable model for image classification. It explains its decisions by providing visual prototypes that resemble specific appearance features. These characteristics are predefined by humans, which on the one hand makes them understandable and on the other hand leads to the model basing its decision on the same features as the human expert. On two public datasets, this method shows better performance compared to existing explainable approaches, despite the additive explainability modality through the visual prototypes. In addition to the performance evaluations, we conducted an analysis of truthfulness by examining the joint information between the target prediction and its explanation output. This was done in order to ensure that the explanation actually reasons the target classification. Through extensive hyperparameter studies, we also found optimal model settings, providing a starting point for further research. Our work emphasises the prospects of combining xAI approaches for greater explainability and demonstrates that incorporating explainability does not necessarily lead to a loss of performance.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2908"},"PeriodicalIF":3.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192993/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Combining convolutional neural network with transformer to improve YOLOv7 for gas plume detection and segmentation in multibeam water column images.","authors":"Wenguang Chen, Xiao Wang, Junjie Chen, Jialong Sun, Guozhen Zha","doi":"10.7717/peerj-cs.2923","DOIUrl":"10.7717/peerj-cs.2923","url":null,"abstract":"<p><p>Multibeam bathymetry has become an effective underwater target detection method by using echo signals to generate a high-resolution water column image (WCI). However, the gas plume in the image is often affected by the seafloor environment and exhibits sparse texture and changing motion, making traditional detection and segmentation methods more time-consuming and labor-intensive. The emergence of convolutional neural networks (CNNs) alleviates this problem, but the local feature extraction of the convolutional operations, while capturing detailed information well, cannot adapt to the elongated morphology of the gas plume target, limiting the improvement of the detection and segmentation accuracy. Inspired by the transformer's ability to achieve global modeling through self-attention, we combine CNN with the transformer to improve the existing YOLOv7 (You Only Look Once version 7) model. First, we sequentially reduce the ELAN (Efficient Layer Aggregation Networks) structure in the backbone network and verify that using the enhanced feature extraction module only in the deep network is more effective in recognising the gas plume targets. Then, the C-BiFormer module is proposed, which can achieve effective collaboration between local feature extraction and global semantic modeling while reducing computing resources, and enhance the multi-scale feature extraction capability of the model. Finally, two different depths of networks are designed by stacking C-BiFormer modules with different numbers of layers. This improves the receptive field so that the model's detection and segmentation accuracy achieve different levels of improvement. Experimental results show that the improved model is smaller in size and more accurate compared to the baseline.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2923"},"PeriodicalIF":3.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192642/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499226","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Gene selection based on adaptive neighborhood-preserving multi-objective particle swarm optimization.","authors":"Sumet Mehta, Fei Han, Muhammad Sohail, Bhekisipho Twala, Asad Ullah, Fasee Ullah, Arfat Ahmad Khan, Qinghua Ling","doi":"10.7717/peerj-cs.2872","DOIUrl":"10.7717/peerj-cs.2872","url":null,"abstract":"<p><p>The analysis of high-dimensional microarray gene expression data presents critical challenges, including excessive dimensionality, increased computational burden, and sensitivity to random initialization. Traditional optimization algorithms often produce inconsistent and suboptimal results, while failing to preserve local data structures limiting both predictive accuracy and biological interpretability. To address these limitations, this study proposes an adaptive neighborhood-preserving multi-objective particle swarm optimization (ANPMOPSO) framework for gene selection. ANPMOPSO introduces four key innovations: (1) a weighted neighborhood-preserving ensemble embedding (WNPEE) technique for dimensionality reduction that retains local structure; (2) Sobol sequence (SS) initialization to enhance population diversity and convergence stability; (3) a differential evolution (DE)-based adaptive velocity update to dynamically balance exploration and exploitation; and (4) a novel ranking strategy that combines Pareto dominance with neighborhood preservation quality to prioritize biologically meaningful gene subsets. Experimental evaluations on six benchmark microarray datasets and eleven multi-modal test functions (MMFs) demonstrate that ANPMOPSO consistently outperforms state-of-the-art methods. For example, it achieves 100% classification accuracy on Leukemia and Small-Round-Blue-Cell Tumor (SRBCT) using only 3-5 genes, improving accuracy by 5-15% over competitors while reducing gene subsets by 40-60%. Additionally, on MMFs, ANPMOPSO attains superior hypervolume values (<i>e.g</i>., 1.0617 ± 0.2225 on MMF1, approximately 10-20% higher than competitors), confirming its robustness in balancing convergence and diversity. Although the method incurs higher training time due to its structural and adaptive components, it achieves a strong trade-off between computational cost and biological relevance, making it a promising tool for high-dimensional gene selection in bioinformatics.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2872"},"PeriodicalIF":3.5,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192825/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning model for gastrointestinal polyp segmentation.","authors":"Zitong Wang, Zeyi Wang, Pengyu Sun","doi":"10.7717/peerj-cs.2924","DOIUrl":"10.7717/peerj-cs.2924","url":null,"abstract":"<p><p>One of the biggest hazards to cancer-related mortality globally is colorectal cancer, and improved patient outcomes are greatly influenced by early identification. Colonoscopy is a highly effective screening method, yet segmentation and detection remain challenging aspects due to the heterogeneity and variability of readers' interpretations of polyps. In this work, we introduce a novel deep learning architecture for gastrointestinal polyp segmentation in the Kvasir-SEG dataset. Our method employs an encoder-decoder structure with a pre-trained ConvNeXt model as the encoder to learn multi-scale feature representations. The feature maps are passed through a ConvNeXt Block and then through a decoder network consisting of three decoder blocks. Our key contribution is the employment of a cross-attention mechanism that creates shortcut connections between the decoder and encoder to maximize feature retention and reduce information loss. In addition, we introduce a Residual Transformer Block in the decoder that learns long-term dependency by using self-attention mechanisms and enhance feature representations. We evaluate our model on the Kvasir-SEG dataset, achieving a Dice coefficient of 0.8715 and mean intersection over union (mIoU) of 0.8021. Our methodology demonstrates state-of-the-art performance in gastrointestinal polyp segmentation and its feasibility of being used as part of clinical pipelines to assist with automated detection and diagnosis of polyps.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2924"},"PeriodicalIF":3.5,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192692/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499240","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Native language identification from text using a fine-tuned GPT-2 model.","authors":"Yuzhe Nie","doi":"10.7717/peerj-cs.2909","DOIUrl":"10.7717/peerj-cs.2909","url":null,"abstract":"<p><p>Native language identification (NLI) is a critical task in computational linguistics, supporting applications such as personalized language learning, forensic analysis, and machine translation. This study investigates the use of a fine-tuned GPT-2 model to enhance NLI accuracy. Using the NLI-PT dataset, we preprocess and fine-tune GPT-2 to classify the native language of learners based on their Portuguese-written texts. Our approach leverages deep learning techniques, including tokenization, embedding extraction, and multi-layer transformer-based classification. Experimental results show that our fine-tuned GPT-2 model significantly outperforms traditional machine learning methods (<i>e.g</i>., SVM, Random Forest) and other pre-trained language models (<i>e.g</i>., BERT, RoBERTa, BioBERT), achieving a weighted F1 score of 0.9419 and an accuracy of 94.65%. These results show that large transformer models work well for native language identification and can help guide future research in personalized language tools and artificial intelligence (AI)-based education.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2909"},"PeriodicalIF":3.5,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12192634/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144499295","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}