PeerJ Computer Science最新文献

{"title":"Activation function cyclically switchable convolutional neural network model.","authors":"İsmail Akgül","doi":"10.7717/peerj-cs.2756","DOIUrl":"10.7717/peerj-cs.2756","url":null,"abstract":"<p><p>Neural networks are a state-of-the-art approach that performs well for many tasks. The activation function (AF) is an important hyperparameter that creates an output against the coming inputs to the neural network model. AF significantly affects the training and performance of the neural network model. Therefore, selecting the most optimal AF for processing input data in neural networks is important. Determining the optimal AF is often a difficult task. To overcome this difficulty, studies on trainable AFs have been carried out in the literature in recent years. This study presents a different approach apart from fixed or trainable AF approaches. For this purpose, the activation function cyclically switchable convolutional neural network (AFCS-CNN) model structure is proposed. The AFCS-CNN model structure does not use a fixed AF value during training. It is designed in a self-regulating model structure by switching the AF during model training. The proposed model structure is based on the logic of starting training with the most optimal AF selection among many AFs and cyclically selecting the next most optimal AF depending on the performance decrease during neural network training. Any convolutional neural network (CNN) model can be easily used in the proposed model structure. In this way, a simple but effective perspective has been presented. In this study, first, ablation studies have been carried out using the Cifar-10 dataset to determine the CNN models to be used in the AFCS-CNN model structure and the specific hyperparameters of the proposed model structure. After the models and hyperparameters were determined, expansion experiments were carried out using different datasets with the proposed model structure. The results showed that the AFCS-CNN model structure achieved state-of-the-art success in many CNN models and different datasets.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2756"},"PeriodicalIF":3.5,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11948315/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143732582","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":"Transformer-based ensemble model for dialectal Arabic sentiment classification.","authors":"Omar Mansour, Eman Aboelela, Remon Talaat, Mahmoud Bustami","doi":"10.7717/peerj-cs.2644","DOIUrl":"10.7717/peerj-cs.2644","url":null,"abstract":"<p><p>Social media platforms such as X, Facebook, and Instagram have become essential avenues for individuals to articulate their opinions, especially during global emergencies. These platforms offer valuable insights that necessitate analysis for informed decision-making and a deeper understanding of societal trends. Sentiment analysis is crucial for assessing public sentiment toward specific issues; however, applying it to dialectal Arabic presents considerable challenges in natural language processing. The complexity arises from the language's intricate semantic and morphological structures, along with the existence of multiple dialects. This form of analysis, also referred to as sentiment classification, opinion mining, emotion mining, and review mining, is the focus of this study, which analyzes tweets from three benchmark datasets: the Arabic Sentiment Tweets Dataset (ASTD), the A Twitter-based Benchmark Arabic Sentiment Analysis Dataset (ASAD), and the Tweets Emoji Arabic Dataset (TEAD). The research involves experimentation with a variety of comparative models, including machine learning, deep learning, transformer-based models, and a transformer-based ensemble model. Feature extraction for both machine learning and deep learning approaches is performed using techniques such as AraVec, FastText, AraBERT, and Term Frequency-Inverse Document Frequency (TF-IDF). The study compares machine learning models such as support vector machine (SVM), naïve Bayes (NB), decision tree (DT), and extreme gradient boosting (XGBoost) with deep learning models such as convolutional neural networks (CNN) and bidirectional long short-term memory (BLSTM) networks. Additionally, it explores transformer-based models such as CAMeLBERT, XLM-RoBERTa, and MARBERT, along with their ensemble configurations. The findings demonstrate that the proposed transformer-based ensemble model achieved superior performance, with average accuracy, recall, precision, and F1-score of 90.4%, 88%, 87.3%, and 87.7%, respectively.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2644"},"PeriodicalIF":3.5,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11948314/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143732585","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":"TurkSentGraphExp: an inherent graph aware explainability framework from pre-trained LLM for Turkish sentiment analysis.","authors":"Yasir Kilic, Cagatay Neftali Tulu","doi":"10.7717/peerj-cs.2729","DOIUrl":"10.7717/peerj-cs.2729","url":null,"abstract":"<p><p>Sentiment classification is a widely studied problem in natural language processing (NLP) that focuses on identifying the sentiment expressed in text and categorizing it into predefined classes, such as positive, negative, or neutral. As sentiment classification solutions are increasingly integrated into real-world applications, such as analyzing customer feedback in business reviews (<i>e.g</i>., hotel reviews) or monitoring public sentiment on social media, the importance of both their accuracy and explainability has become widely acknowledged. In the Turkish language, this problem becomes more challenging due to the complex agglutinative structure of the language. Many solutions have been proposed in the literature to solve this problem. However, it is observed that the solutions are generally based on black-box models. Therefore the explainability requirement of such artificial intelligence (AI) models has become as important as the accuracy of the model. This has further increased the importance of studies based on the explainability of the AI model's decision. Although most existing studies prefer to explain the model decision in terms of the importance of a single feature/token, this does not provide full explainability due to the complex lexical and semantic relations in the texts. To fill these gaps in the Turkish NLP literature, in this article, we propose a graph-aware explainability solution for Turkish sentiment analysis named TurkSentGraphExp. The solution provides both classification and explainability for sentiment classification of Turkish texts by considering the semantic structure of suffixes, accommodating the agglutinative nature of Turkish, and capturing complex relationships through graph representations. Unlike traditional black-box learning models, this framework leverages an inherent graph representation learning (GRL) model to introduce rational phrase-level explainability. We conduct several experiments to quantify the effectiveness of this framework. The experimental results indicate that the proposed model achieves a 10 to 40% improvement in explainability compared to state-of-the-art methods across varying sparsity levels, further highlighting its effectiveness and robustness. Moreover, the experimental results, supported by a case study, reveal that the semantic relationships arising from affixes in Turkish texts can be identified as part of the model's decision-making process, demonstrating the proposed solution's ability to effectively capture the agglutinative structure of Turkish.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2729"},"PeriodicalIF":3.5,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935768/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712124","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":"Application of artificial intelligence technology in the economic development of urban intelligent transportation system.","authors":"Ziming Zhao, Jinyu Chen","doi":"10.7717/peerj-cs.2728","DOIUrl":"https://doi.org/10.7717/peerj-cs.2728","url":null,"abstract":"<p><p>With the rapid development of the social economy and the gradual improvement of residents' living standards, the increasing number of urban cars has exacerbated urban traffic congestion. This article analyzed the application of artificial intelligence (AI) technology in five aspects of urban intelligent transportation systems. Artificial intelligence technology was used in traffic data collection and processing to provide accurate data support for traffic decision-making. A traffic flow prediction model was established for traffic flow prediction and optimized scheduling algorithms were used to dispatch vehicles on congested urban roads intelligently. Artificial intelligence algorithms can be used to optimize urban traffic signal control systems in intelligent traffic signal control; artificial intelligence technology can be applied to develop intelligent driving systems in the fields of intelligent driving and traffic safety; in terms of data analysis and decision support, it can use AI technology to analyze a large number of traffic data to provide decision support for urban traffic managers, and analyze the impact of the application of AI technology in urban intelligent transportation system on urban economic growth. This article evaluated the economic benefits of artificial intelligence technology in urban intelligent transportation systems. The evaluation results show that the total economic cost of the urban intelligent transportation system after the application of AI technology was 2,961 yuan less than before the application of AI technology, significantly reducing the investment cost of roads. This article analyzes the application of artificial intelligence technology in the economic development of intelligent urban transportation systems, which can meet the needs of healthy urban development and ensure road traffic safety.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2728"},"PeriodicalIF":3.5,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935779/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712146","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":"Proactive detection of anomalous behavior in Ethereum accounts using XAI-enabled ensemble stacking with Bayesian optimization.","authors":"Vasavi Chithanuru, Mangayarkarasi Ramaiah","doi":"10.7717/peerj-cs.2630","DOIUrl":"10.7717/peerj-cs.2630","url":null,"abstract":"<p><p>The decentralized, open-source architecture of blockchain technology, exemplified by the Ethereum platform, has transformed online transactions by enabling secure and transparent exchanges. However, this architecture also exposes the network to various security threats that cyber attackers can exploit. Detecting suspicious behaviors in account on the Ethereum blockchain can help mitigate attacks, including phishing, Ponzi schemes, eclipse attacks, Sybil attacks, and distributed denial of service (DDoS) incidents. The proposed system introduces an ensemble stacking model combining Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and a neural network (NN) to detect potential threats within the Ethereum platform. The ensemble model is fine-tuned using Bayesian optimization to enhance predictive accuracy, while explainable artificial intelligence (XAI) tools-SHAP, LIME, and ELI5-provide interpretable feature insights, improving transparency in model predictions. The dataset used comprises 9,841 Ethereum transactions across 52 initial fields (reduced to 17 relevant features), encompassing both legitimate and fraudulent records. The experimental findings demonstrate that the proposed model achieves a superior accuracy of 99.6%, outperforming that of other cutting-edge methods. These findings demonstrate that the XAI-enabled ensemble stacking model offers a highly effective, interpretable solution for blockchain security, strengthening trust and reliability within the Ethereum ecosystem.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2630"},"PeriodicalIF":3.5,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935758/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712110","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":"Protein-protein interaction prediction using enhanced features with spaced conjoint triad and amino acid pairwise distance.","authors":"Yunus Emre Göktepe","doi":"10.7717/peerj-cs.2748","DOIUrl":"10.7717/peerj-cs.2748","url":null,"abstract":"<p><p>Protein-protein interactions (PPIs) are pivotal in cellular processes, influencing a wide range of functions, from metabolism to immune responses. Despite the advancements in experimental techniques for PPI detection, their inherent limitations, such as high false-positive rates and significant resource demands, necessitate the development of computational approaches. This study presents a novel computational model named MFPIC (Multi-Feature Protein Interaction Classifier) for predicting PPIs, integrating enhanced sequence-based features, including a novel spaced conjoint triad (SCT) and amino acid pairwise distance (AAPD), with existing methods such as position-specific scoring matrices (PSSM) and AAindex-based features. The SCT captures complex sequence motifs by considering non-adjacent amino acid interactions, while AAPD provides critical spatial information about amino acid residues within protein sequences. The proposed model was evaluated across three benchmark datasets-<i>Saccharomyces cerevisiae</i>, <i>Helicobacter pylori</i>, and human proteins-demonstrating superior performance in comparison to state-of-the-art models. The results underscore the efficacy of integrating diverse and complementary features, achieving significant improvements in predictive accuracy, with the model achieving 95.90%, 99.33%, and 90.95% accuracy on the <i>Saccharomyces cerevisiae</i>, <i>Helicobacter pylori</i>, and human dataset, respectively. This approach not only enhances our understanding of PPI mechanisms but also offers valuable insights for the development of targeted therapeutic strategies.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2748"},"PeriodicalIF":3.5,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935777/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712149","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":"Graph-based deep fusion for architectural text representation.","authors":"Shaoyun Hu, Qingxiong Weng","doi":"10.7717/peerj-cs.2735","DOIUrl":"10.7717/peerj-cs.2735","url":null,"abstract":"<p><p>Amidst the swift global urbanization and rapid evolution of the architecture industry, there is a growing demand for the automated processing of architectural textual information. This demand arises from the abundance of specialized vocabulary in architectural texts, posing a challenge for accurate representation using traditional models. To address this, we propose a novel fusion method that integrates Transformer-based models with graph neural networks (GNNs) for architectural text representation. While independently utilizing Bidirectional Encoder Representations from Transformers (BERT) and the robustly optimized BERT approach (RoBERTa) to generate initial document representations, we also employ term frequency-inverse document frequency (TF-IDF) to extract keywords from each document and construct a corresponding keyword set. Subsequently, a graph is created based on the keyword vocabulary and document embeddings, which is then fed into the graph attention network (GAT). The final document embedding is generated by GAT, and the text embedding is crafted by the attention module and neural network structure of the GAT. Experimental results from comparison studies show that the proposed model outperforms all baselines. Additionally, ablation studies demonstrate the effectiveness of each module, further reinforcing the robustness and superiority of our approach.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2735"},"PeriodicalIF":3.5,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935773/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712094","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 line follower robot: a meta-analytic approach.","authors":"Williamson Johnny Hatzinakis Brigido, Jose M Parente de Oliveira","doi":"10.7717/peerj-cs.2744","DOIUrl":"10.7717/peerj-cs.2744","url":null,"abstract":"<p><p>Line-follower robots represent a critical segment in autonomous robotics, with broad applications ranging from industrial automation to educational tools. This meta-analytic review synthesizes research on line-follower robots, addressing a noticeable gap in the literature where comprehensive analyses are scarce. The review leverages the Theory of the Consolidated Meta-analytic Approach (TEMAC) to systematically explore 287 documents spanning from 2001 to 2024, highlighting key contributions, trends, and gaps in the field. Through this analysis, it becomes evident that while significant advancements have been made in control strategies, sensor integration, and noise reduction techniques, the literature still lacks comprehensive studies on the scalability of these technologies, especially in large-scale industrial environments. Recent research trends emphasize integrating artificial intelligence and machine learning into line-follower robots, indicating a shift towards more sophisticated, adaptable systems. Despite these advancements, challenges remain in addressing environmental variability, improving real-time adaptability, and exploring novel applications in dynamic environments. This review not only maps the historical evolution and current state of line-follower robots but also identifies future research directions that could drive the next generation of robotic systems. The findings offer valuable insights for researchers, engineers, and educators aiming to enhance the efficiency, reliability, and application scope of line-follower robots.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2744"},"PeriodicalIF":3.5,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935780/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712123","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":"Protein language model-based prediction for plant miRNA encoded peptides.","authors":"Yishan Yue, Henghui Fan, Jianping Zhao, Junfeng Xia","doi":"10.7717/peerj-cs.2733","DOIUrl":"10.7717/peerj-cs.2733","url":null,"abstract":"<p><p>Plant miRNA encoded peptides (miPEPs), which are short peptides derived from small open reading frames within primary miRNAs, play a crucial role in regulating diverse plant traits. Plant miPEPs identification is challenging due to limitations in the available number of known miPEPs for training. Existing prediction methods rely on manually encoded features, including miPEPPred-FRL, to infer plant miPEPs. Recent advances in deep learning modeling of protein sequences provide an opportunity to improve the representation of key features, leveraging large datasets of protein sequences. In this study, we propose an accurate prediction model, called pLM4PEP, which integrates ESM2 peptide embedding with machine learning methods. Our model not only demonstrates precise identification capabilities for plant miPEPs, but also achieves remarkable results across diverse datasets that include other bioactive peptides. The source codes, datasets of pLM4PEP are available at https://github.com/xialab-ahu/pLM4PEP.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2733"},"PeriodicalIF":3.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935769/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712147","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 cassava disease recognition using object segmentation and progressive learning.","authors":"Chang Che, Nian Xue, Zhen Li, Yilin Zhao, Xin Huang","doi":"10.7717/peerj-cs.2721","DOIUrl":"10.7717/peerj-cs.2721","url":null,"abstract":"<p><p>Cassava is a vital crop for millions of farmers worldwide, but its cultivation is threatened by various destructive diseases. Current detection methods for cassava diseases are costly, time-consuming, and often limited to controlled environments, making them unsuitable for large-scale agricultural use. This study aims to develop a deep learning framework that enables early, accurate, and efficient detection of cassava diseases in real-world conditions. We propose a self-supervised object segmentation technique, combined with a progressive learning algorithm (PLA) that incorporates both triplet loss and classification loss to learn robust feature embeddings. Our approach achieves superior performance on the Cassava Leaf Disease Classification (CLDC) dataset from the Kaggle competition, with an accuracy of 91.43%, outperforming all other participants. The proposed method offers a practical and efficient solution for cassava disease detection, demonstrating the potential for large-scale, real-world application in agriculture.</p>","PeriodicalId":54224,"journal":{"name":"PeerJ Computer Science","volume":"11 ","pages":"e2721"},"PeriodicalIF":3.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11935765/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143712150","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}