Neurocomputing最新文献

{"title":"A hybrid approach combining sentiment analysis and deep learning to mitigate data sparsity in recommender systems","authors":"Ikram Karabila ,&nbsp;Nossayba Darraz ,&nbsp;Anas El-Ansari ,&nbsp;Nabil Alami ,&nbsp;Mostafa El Mallahi","doi":"10.1016/j.neucom.2025.129886","DOIUrl":"10.1016/j.neucom.2025.129886","url":null,"abstract":"<div><div>The optimization of recommendation systems (RS) is crucial for delivering personalized product suggestions. Despite their successes, RS approaches often face challenges, such as data sparsity in the user–item matrix, which can undermine their performance. To address these challenges, integrating additional information sources, such as item/user profiles and textual reviews, is essential. These sources offer valuable insights into user preferences and item characteristics, helping in understanding the contextual details of both. This study focuses on developing an advanced RS architecture that combines Singular Value Decomposition (SVD) with BERT-CB methods and a Hybrid Model-based Sentiment Analysis. By integrating BERT with Multilayer Perceptron (MLP) methods, the system gains a deeper understanding of item profiles, improving the comprehension of user preferences and item characteristics. Additionally, a novel hybrid approach for sentiment analysis is proposed, using GloVe embeddings and CNN-BiGRU, improving the accuracy and robustness of sentiment detection in user reviews. This comprehensive understanding, combined with collaborative filtering models like SVD, enables the system to provide highly accurate recommendations. The proposed approach consists of four main phases: first, embedding review text using GloVe embeddings and developing a hybrid sentiment analysis approach with CNN and BiGRU architectures; second, creating a BERT language model for generating embeddings from item profile texts, followed by dimensionality reduction using Auto-Encoder; third, using these vectors to build a novel MLP model; fourth, developing a Collaborative Filtering method using SVD, and finally, combining these methods into a hybrid approach and conducting a comprehensive evaluation. Empirical results clearly show the effectiveness of our approach, particularly the combination of GloVe-CNN-BiGRU and BERT-CB with SVD methodology, demonstrating significant improvements across various performance metrics. This confirms the practical value of using contextualized data from BERT-CB and the sentiment analysis approach, enhancing the recommendation system’s effectiveness.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129886"},"PeriodicalIF":5.5,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143686989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A reformulation neurodynamic algorithm for distributed nonconvex optimization","authors":"Xin Yu,&nbsp;Qingzhou Huang,&nbsp;Rixin Lin","doi":"10.1016/j.neucom.2025.130023","DOIUrl":"10.1016/j.neucom.2025.130023","url":null,"abstract":"<div><div>This paper presents a reformulation neurodynamic algorithm for solving distributed nonconvex optimization problems. A class of general Lagrangian functions is introduced to eliminate the dual gap in nonconvex problems. This algorithm extends the application of neurodynamic algorithms based on the <span><math><mi>p</mi></math></span>-power reformulation transformation of Lagrangian functions. Under mild conditions, the initial point of the decision vector can be arbitrarily chosen. It is proven that the output trajectories will eventually converge to a strict local minimum point of the distributed nonconvex optimization problem. Finally, numerical experiments demonstrate the effectiveness of the proposed algorithm, which is also applied to solve the oblique throwing problem and the distributed source localization problem.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"635 ","pages":"Article 130023"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143643977","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fine-grained hierarchical singular value decomposition for convolutional neural networks compression and acceleration","authors":"Mengmeng Qi ,&nbsp;Dingheng Wang ,&nbsp;Wei Yang ,&nbsp;Baorong Liu ,&nbsp;Fuyong Wang ,&nbsp;Zengqiang Chen","doi":"10.1016/j.neucom.2025.129966","DOIUrl":"10.1016/j.neucom.2025.129966","url":null,"abstract":"<div><div>Convolutional neural networks (CNNs) still remain crucial in the field of computer vision, especially in industrial-embedded scenarios. Although modern artificial intelligence chips such as embedded graphics processing units (GPUs) and neural process units (NPUs) are equipped with sufficient computability, making CNNs more lightweight always has non-negligible significance. Until now, many researchers have made multiple corresponding achievements, in which a series of tensor decomposition methods have represented their unique advantages such as concision, flexibility, and low-rank approximation theory. However, balancing the compression, acceleration, and precision, is still an open issue, because the traditional tensor decompositions are hard to deal with the trade-off between approximation and compression ability, while the so-called fine-grained tensor decompositions such as Kronecker canonical polyadic (KCP) have not created a way to merge the factors for efficient inference. In this paper, we first review related works on convolutional neural network (CNN) compression and the necessary prior knowledge. We then propose a novel matrix decomposition method, termed hierarchical singular value (HSV) decomposition, and validate its effectiveness. Subsequently, we introduce a fast contraction strategy based on the merged factors of HSV and explain how our method addresses the inefficiencies in inference associated with traditional contraction processes. Additionally, we validate the advantages of HSV by comparing its complexity with that of other classical tensor decomposition methods. Thereafter, we apply HSV to CNN compression and acceleration by transforming convolution operations into matrix multiplication. We also propose a self-adaptive rank selection algorithm tailored to standard CNN architecture and conduct a theoretical analysis of the convergence of our method. Multiple experiments on CIFAR-10, ImageNet, COCO, and Cityscapes benchmark datasets show that the proposed HSV-Conv can simultaneously gain considerable compression ratio and acceleration ratio, while the precision loss is almost non-existent. We also make a comprehensive comparison with the other related works, and the superiority of our method is further validated. Besides, we give a deep discussion about the rank selection issue of HSV in the aspects of practice and theory, which explains the strategy of the proposed self-adaptive rank selection and the reason for choosing fine-tuning rather than training from scratch.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129966"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143695956","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Privacy-preserving and Byzantine-robust federated broad learning with chain-loop structure","authors":"Nan Li,&nbsp;Chang-E Ren,&nbsp;Siyao Cheng","doi":"10.1016/j.neucom.2025.129975","DOIUrl":"10.1016/j.neucom.2025.129975","url":null,"abstract":"<div><div>Federated learning (FL) can collaboratively train a model by aggregating local models instead of aggregating raw data, which can protect privacy by ensuring that data remains on the client. However, the traditional FL still faces some challenges such as privacy leakage and the presence of Byzantine clients. We propose a privacy-preserving and Byzantine-robust federated broad learning framework with chain-loop structure i.e., PBFBL-CL, and this algorithm can simultaneously achieve protection of clients’ privacy and robustness against Byzantine attacks. In this paper, we apply Byzantine step-by-step co-validation algorithm to address the existence of Byzantine clients. We pass the aggregated model through the chain, so each client’s privacy is well protected. Moreover, PBFBL-CL can reduce the communication overhead between clients and server. Finally, we evaluate the PBFBL-CL algorithm in MNIST, Fashion-MNIST and NORB datasets, and the results show that our algorithm is better than existing FL algorithms in terms of model accuracy and training speed. Experimental results demonstrate that under the extreme scenario where Byzantine client proportion reaches 90%, the model achieves an accuracy of 89.53%, only 4.17% lower than the 93.7% accuracy observed in the ideal scenario without Byzantine clients.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129975"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143686982","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adjustable behavior-guided adaptive dynamic programming for neural learning control","authors":"Guohan Tang,&nbsp;Ding Wang,&nbsp;Ao Liu,&nbsp;Junfei Qiao","doi":"10.1016/j.neucom.2025.129986","DOIUrl":"10.1016/j.neucom.2025.129986","url":null,"abstract":"<div><div>In this article, an adjustable behavior-guided adaptive dynamic programming (BGADP) algorithm is designed to solve the optimal regulation problem for discrete-time systems. In conventional adaptive dynamic programming methods, gradient information of system dynamics is necessary for conducting policy improvement. However, these methods face challenges when gradient information cannot be computed or when the system dynamics is non-differentiable. To overcome these limitations, a human-behavior-inspired swarm intelligence approach is used to search for superior policies during the iterative process, eliminating the need for gradient information. Additionally, a relaxation factor is introduced into the value function update to accelerate the convergence speed of the algorithm. The monotonicity and convergence properties of the iterative value function are rigorously analyzed. Finally, the effectiveness and practicality of the adjustable BGADP algorithm are validated through two simulation studies, which are implemented using the actor–critic framework with neural networks.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129986"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Student behavior detection model based on multilevel residual networks and hybrid attention mechanisms","authors":"Wenbin Lu,&nbsp;Songyan Liu,&nbsp;Boyang Ding,&nbsp;Peng Chen,&nbsp;Fangpeng Lu","doi":"10.1016/j.neucom.2025.129965","DOIUrl":"10.1016/j.neucom.2025.129965","url":null,"abstract":"<div><div>Accurately analyzing student behaviors allows for better evaluation of student engagement, which in turn can improve teaching quality. To address challenges such as multi-scale scenes, occluded targets, and subtle fine features in classroom environments, while also considering model implementability, we propose an efficient student behavior detection model, RSAY. This model leverages multi-scale information extraction and a hybrid attention mechanism to support teaching. Both the backbone and feature fusion networks of the model integrate our designed Rep_SC_Atten module, which incorporates our novel multi-level residual network architecture and a lightweight hybrid attention mechanism. This hybrid architecture enhances the model’s sensitivity and ability to extract multi-scale information, while ensuring effective extraction of fine-grained features via the attention mechanism. Additionally, the DDetect strategy is introduced in the detection head to reduce model size without sacrificing accuracy. We evaluated our model using the SCB-Dataset and a custom student behavior dataset, demonstrating a 6.3% improvement in accuracy over the baseline model.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"635 ","pages":"Article 129965"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143644086","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spectrum-guided Spatial Feature Enhancement Network for event-based lip-reading","authors":"Yi Zhang ,&nbsp;Xiuping Liu ,&nbsp;Hongchen Tan ,&nbsp;Xin Li","doi":"10.1016/j.neucom.2025.129974","DOIUrl":"10.1016/j.neucom.2025.129974","url":null,"abstract":"<div><div>The Automatic Lip-reading task aims to recognize spoken words through visual cues from the speaker’s lip movements. This crucial task complements audio-based speech recognition systems and can substitute them when sound is unavailable. Event-based lip-reading methods have gained increasing attention due to the advantages of event cameras, such as high temporal resolution and low power consumption. However, existing methods often fail to fully utilize the spatial information in event data due to its sparsity and the presence of random activations. To address this, we propose a novel Spectral-guided Spatial Enhancement Network (SSE-Net). SSE-Net introduces two core innovations: the Spectrum-guided Spatial Feature Enhance Module (SSEM) and the Multi-Scale Spatial Interaction Module (MS-SIM). SSEM employs frequency domain enhancement and spatial feature enhancement strategies to augment spatial features crucial for event-based lipreading tasks. MS-SIM conducts the fusion and interaction of multi-level semantics, enriching the contextual information of lip representations. We conducted experiments on the event-based lip-reading dataset DVS-Lip with our proposed method and demonstrated its superiority over other state-of-the-art event-based lip-reading methods.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129974"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687917","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Texture dominated no-reference quality assessment for high resolution image by multi-scale mechanism","authors":"Ziqing Huang ,&nbsp;Hao Liu ,&nbsp;Zhihao Jia ,&nbsp;Shuo Zhang ,&nbsp;Yonghua Zhang ,&nbsp;Shiguang Liu","doi":"10.1016/j.neucom.2025.130003","DOIUrl":"10.1016/j.neucom.2025.130003","url":null,"abstract":"<div><div>With the rapid development of new media formats, various high-definition display devices are ubiquitous, and high-resolution (HR) images are essential for high-quality visual experiences. Quality assessment of HR images has become an urgent challenge. However, conventional image quality assessment (IQA) methods with good performance are designed for low-resolution (LR) images, which lacks the perceptual characteristics of HR images, resulting in difficult to achieve satisfactory subjective consistency. Moreover, huge computational costs would have to be consumed when applying those deep neural networks in LR-IQA directly to HR images. Inspired by the fact that regions with rich textures are more sensitive to distortion than others, texture dominated no-reference image quality assessment for HR images are proposed in this paper. Specifically, a dual branch network based on multi-scale technology was designed to extract texture and semantic features separately, and cross scale and dual dimensional attention were introduced to ensure the dominance of texture features. Then, multi-layer perception network is used to map the extracted quality perception feature vectors to the predicted quality score. Worthy of note is that local entropy has been calculated and representative blocks are cropped as inputs to the feature extraction network, greatly reducing computational complexity. Overall, the texture dominated high-resolution IQA network (TD-HRNet) proposed utilizes a reference free method, while could perform excellently on HR datasets of different sizes, image types, and distortion types, accurately predicting the quality of different types of HR images.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 130003"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143706021","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Channel pruning for convolutional neural networks using l0-norm constraints","authors":"Enhao Chen ,&nbsp;Hao Wang ,&nbsp;Zhanglei Shi ,&nbsp;Wei Zhang","doi":"10.1016/j.neucom.2025.129925","DOIUrl":"10.1016/j.neucom.2025.129925","url":null,"abstract":"<div><div>Channel pruning can effectively reduce the size and inference time of Convolutional Neural Networks (CNNs). However, existing channel pruning methods still face several issues, including high computational costs, extensive manual intervention, difficulty in hyperparameter tuning, and challenges in directly controlling the sparsity. To address these issues, this paper proposes two channel pruning methods based on <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm sparse optimization: the <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm Pruner and the Automated <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm Pruner. The <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm Pruner formulates the channel pruning problem as a sparse optimization problem involving the <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm and achieves a fast solution through a series of approximations and transformations. Inspired by this solution process, we devise the Zero-Norm (ZN) module, which can autonomously select output channels for each layer based on a predefined global pruning ratio. This approach incurs low computational cost and allows for precise control over the overall pruning ratio. Furthermore, to further enhance the performance of the pruned model, we have developed the Automated <span><math><msub><mrow><mi>l</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>-norm Pruner. This method utilizes a Bee Colony Optimization algorithm to adjust the pruning ratio, mitigating the negative impact of manually preset pruning ratios on model performance. Our experiments demonstrate that the proposed pruning methods outperform several state-of-the-art techniques. The source code for our proposed methods is available at: <span><span>https://github.com/TCCofWANG/l0_prune</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129925"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687916","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Parallel Spatiotemporal Network to recognize micro-expression","authors":"Jingting Li ,&nbsp;Su-Jing Wang ,&nbsp;Yong Wang ,&nbsp;Haoliang Zhou ,&nbsp;Xiaolan Fu","doi":"10.1016/j.neucom.2025.129891","DOIUrl":"10.1016/j.neucom.2025.129891","url":null,"abstract":"<div><div>Micro-expressions are fleeting spontaneous facial expressions that commonly occur in high-stakes scenarios and reflect humans’ mental states. Thus, it is one of the crucial clues for lie detection. Furthermore, due to the brief duration of micro-expression, temporal information is important for micro-expression recognition. The paper proposes a Parallel Spatiotemporal Network (PSN) to recognize micro-expression. The proposed PSN includes a spatial sub-network and a temporal sub-network. The spatial sub-network is a shallow network with subtle motion information as the input. And the temporal sub-network is a network with a novel temporal feature extraction unit that extracts sparse temporal features of micro-expressions. Finally, we propose an element-wise addition with 1 × 1 convolutional kernel fusion model to fuse the spatial and temporal features. The proposed PSN gets better measurement metrics (such as recognition rate, F1 score, true positive rate, and true negative rate) than the other state-of-the-art methods on the consisted databases consisting of CASME, CASME II, CAS(ME)<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span>, and SAMM.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"636 ","pages":"Article 129891"},"PeriodicalIF":5.5,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}