Neural Networks最新文献_第5页

Anatomical connectivity reconstruction of biological neuronal networks using Granger causality. 基于格兰杰因果关系的生物神经网络解剖连通性重建。

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-08 DOI: 10.1016/j.neunet.2025.108528

Bo Wang, Kai Chen, Shouwei Luo, Yanyang Xiao, Songting Li, Douglas Zhou

Accessing to the anatomical connectivity of the cortical network is crucial for understanding the dynamics and functions of the brain. While direct experimental measurements of anatomical connectivity are costly, effective connectivity methods offer a potential alternative approach to reconstructing anatomical connections with the help of statistical tools. Granger causality (GC) is one of the most widely used tools for network effective connection estimations in brain networks. However, whether the effective connectivity estimated by GC can help to reliably capture the information of the underlying anatomical connectivity of neuronal networks remains largely unknown. In this work, we demonstrate that GC can effectively reconstruct the anatomical connectivity of Hodgkin-Huxley (HH) neuronal networks using neuronal voltage time series data across various dynamical regimes. Moreover, we uncover the quantitative mechanisms underlying the accurate reconstruction capabilities of GC. Furthermore, we extend our analysis from HH type point neuronal networks to multi-compartment neuronal networks, and from voltage data to spike-train data. The GC-based reconstruction remains consistently effective across these different scenarios. Finally, we investigate GC-based reconstruction using real experimental data from Allen Institute, demonstrating that GC-reconstructed connectivities exhibit high consistency across different stimulus conditions. Overall, our findings provide a strong theoretical foundation for the use of GC in realistic neuronal network reconstructions.

获取皮层网络的解剖学连通性对于理解大脑的动力学和功能至关重要。虽然解剖连接的直接实验测量是昂贵的，但有效的连接方法提供了一种潜在的替代方法，可以在统计工具的帮助下重建解剖连接。格兰杰因果关系（Granger causality， GC）是脑网络中使用最广泛的网络有效连接估计工具之一。然而，GC估计的有效连通性是否有助于可靠地捕获神经网络潜在解剖连通性的信息，在很大程度上仍然未知。在这项工作中，我们证明了GC可以利用不同动态状态下的神经元电压时间序列数据有效地重建霍奇金-赫胥黎（HH）神经元网络的解剖连通性。此外，我们还揭示了GC精确重建能力的定量机制。此外，我们将HH型点神经网络的分析扩展到多室神经网络，并将电压数据扩展到尖峰序列数据。基于gc的重建在这些不同的场景中始终保持有效。最后，我们利用Allen研究所的真实实验数据研究了基于gc重建的连接性，结果表明gc重建的连接性在不同的刺激条件下表现出高度的一致性。总的来说，我们的发现为GC在现实神经网络重建中的应用提供了强有力的理论基础。

{"title":"Anatomical connectivity reconstruction of biological neuronal networks using Granger causality.","authors":"Bo Wang, Kai Chen, Shouwei Luo, Yanyang Xiao, Songting Li, Douglas Zhou","doi":"10.1016/j.neunet.2025.108528","DOIUrl":"https://doi.org/10.1016/j.neunet.2025.108528","url":null,"abstract":"<p><p>Accessing to the anatomical connectivity of the cortical network is crucial for understanding the dynamics and functions of the brain. While direct experimental measurements of anatomical connectivity are costly, effective connectivity methods offer a potential alternative approach to reconstructing anatomical connections with the help of statistical tools. Granger causality (GC) is one of the most widely used tools for network effective connection estimations in brain networks. However, whether the effective connectivity estimated by GC can help to reliably capture the information of the underlying anatomical connectivity of neuronal networks remains largely unknown. In this work, we demonstrate that GC can effectively reconstruct the anatomical connectivity of Hodgkin-Huxley (HH) neuronal networks using neuronal voltage time series data across various dynamical regimes. Moreover, we uncover the quantitative mechanisms underlying the accurate reconstruction capabilities of GC. Furthermore, we extend our analysis from HH type point neuronal networks to multi-compartment neuronal networks, and from voltage data to spike-train data. The GC-based reconstruction remains consistently effective across these different scenarios. Finally, we investigate GC-based reconstruction using real experimental data from Allen Institute, demonstrating that GC-reconstructed connectivities exhibit high consistency across different stimulus conditions. Overall, our findings provide a strong theoretical foundation for the use of GC in realistic neuronal network reconstructions.</p>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"108528"},"PeriodicalIF":6.3,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146013046","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Temporal local attention with adaptive decoding: Enhancing spiking neural networks for temporal computing applications 具有自适应解码的时间局部注意：用于时间计算应用的增强尖峰神经网络

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-08 DOI: 10.1016/j.neunet.2026.108558

Hanxiao Fan , Hanle Zheng , Zikai Wang , Jiayi Mao , Huifeng Yin , Hao Guo , Lei Deng

The brain-inspired spiking neural networks (SNNs) are considered to have great potential in complex learning due to their rich neural dynamics and high energy efficiency. Their unique mechanisms are naturally suited for performing temporal computing tasks. However, whether SNNs can effectively capture sufficient temporal features solely based on the neural dynamics remains to be explored, as they usually encounter difficulties in long-sequence gradient propagation during training. In this work, we introduce temporal local attention (TLA), which helps SNNs effectively reduce the sequence length, thereby enhancing the model performance and accelerating the training process. Furthermore, we optimize the output process of SNNs by incorporating adaptive decoding (AD) based on finding the correlation between the decoding strategy and model performance. Finally, we combine these two mechanisms (TLA-AD) into complete SNN modeling with training and inference. We use LSTM and a recent SNN model with dendritic heterogeneity (DH-SNN) as baselines on Electroencephalogram (EEG) and natural language processing (NLP) datasets. The experimental results demonstrate that the proposed TLA-AD method significantly enhances the performance of SNNs and accelerates the training without a significant increase of the number of parameters. It achieves state-of-the-art accuracy scores of 93.52% and 93.41% on the DEAP dataset (valence and arousal), as well as competitive accuracy scores of 87.41% and 86.31% on SEED and IMDB datasets, respectively, compared to other SNN methods. This work provides effective optimization approaches for enhancing SNNs with advanced performance in temporal computing tasks.

大脑激发的脉冲神经网络（SNNs）由于其丰富的神经动力学和高能量效率而被认为在复杂学习中具有巨大的潜力。它们独特的机制自然适合于执行时间计算任务。然而，由于snn在训练过程中通常会遇到长序列梯度传播的困难，因此仅基于神经动力学的snn能否有效捕获足够的时间特征还有待探索。在这项工作中，我们引入了时间局部注意（temporal local attention， TLA），它可以帮助snn有效地减少序列长度，从而提高模型性能并加速训练过程。此外，我们在发现解码策略与模型性能之间的相关性的基础上，通过引入自适应解码（AD）来优化snn的输出过程。最后，我们将这两种机制（TLA-AD）结合起来，形成完整的SNN模型，并进行训练和推理。我们使用LSTM和最近的具有树突异质性的SNN模型（DH-SNN）作为脑电图（EEG）和自然语言处理（NLP）数据集的基线。实验结果表明，提出的TLA-AD方法在不显著增加参数数量的情况下，显著提高了snn的性能，加快了训练速度。与其他SNN方法相比，该方法在DEAP数据集（效价和唤醒）上的准确率分别为93.52%和93.41%，在SEED和IMDB数据集上的竞争准确率分别为87.41%和86.31%。这项工作为增强snn在时间计算任务中的高级性能提供了有效的优化方法。

{"title":"Temporal local attention with adaptive decoding: Enhancing spiking neural networks for temporal computing applications","authors":"Hanxiao Fan , Hanle Zheng , Zikai Wang , Jiayi Mao , Huifeng Yin , Hao Guo , Lei Deng","doi":"10.1016/j.neunet.2026.108558","DOIUrl":"10.1016/j.neunet.2026.108558","url":null,"abstract":"<div><div>The brain-inspired spiking neural networks (SNNs) are considered to have great potential in complex learning due to their rich neural dynamics and high energy efficiency. Their unique mechanisms are naturally suited for performing temporal computing tasks. However, whether SNNs can effectively capture sufficient temporal features solely based on the neural dynamics remains to be explored, as they usually encounter difficulties in long-sequence gradient propagation during training. In this work, we introduce temporal local attention (TLA), which helps SNNs effectively reduce the sequence length, thereby enhancing the model performance and accelerating the training process. Furthermore, we optimize the output process of SNNs by incorporating adaptive decoding (AD) based on finding the correlation between the decoding strategy and model performance. Finally, we combine these two mechanisms (TLA-AD) into complete SNN modeling with training and inference. We use LSTM and a recent SNN model with dendritic heterogeneity (DH-SNN) as baselines on Electroencephalogram (EEG) and natural language processing (NLP) datasets. The experimental results demonstrate that the proposed TLA-AD method significantly enhances the performance of SNNs and accelerates the training without a significant increase of the number of parameters. It achieves state-of-the-art accuracy scores of 93.52% and 93.41% on the DEAP dataset (valence and arousal), as well as competitive accuracy scores of 87.41% and 86.31% on SEED and IMDB datasets, respectively, compared to other SNN methods. This work provides effective optimization approaches for enhancing SNNs with advanced performance in temporal computing tasks.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"Article 108558"},"PeriodicalIF":6.3,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Kernelized linear principal component discriminant analysis 核化线性主成分判别分析。

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-07 DOI: 10.1016/j.neunet.2026.108539

Lingxiao Qu , Yan Pei

In this paper, we propose Kernelized Linear Principal Component Discriminant Analysis (KLPCDA), a structured and unified framework for discriminant analysis that overcomes the fragmentation in existing multi-stage approaches such as PCA+LDA or KPCA+GDA. Instead of treating feature extraction and class discrimination as disjoint steps, KLPCDA formulates a joint optimization model in the Reproducing Kernel Hilbert Space (RKHS), integrating overall variance preservation, between-class separation, and within-class compactness into a fused objective. The formulation supports seven KLPCDA variants, offering flexible control over effects of each criterion through tunable fusion coefficients. We present a systematic parameter optimization strategy, including kernel parameter selection, subspace dimensionality tuning, and fusion balancing, along with an alternative kernel parameter optimization method. Extensive experiments across image, tabular, and signal datasets across small-sample-size (SSS) to larger-scale settings validate the adaptability of KLPCDA. The results demonstrate KLPCDA consistently outperforms benchmark methods and convolutional neural networks in SSS settings on both recognition accuracy and efficiency, while maintaining competitive advantages in computational complexity and storage requirements for large-scale scenarios. Finally, we provide insights into extended study subjects and future work related to our proposal.

本文提出了一种结构化、统一的判别分析框架——核化线性主成分判别分析（KLPCDA），克服了PCA+LDA或KPCA+GDA等多阶段判别分析方法的碎片化问题。KLPCDA没有将特征提取和类判别作为不相关的步骤，而是在再现核希尔伯特空间（RKHS）中建立了一个联合优化模型，将总体方差保持、类间分离和类内紧密性整合到一个融合目标中。该配方支持七个KLPCDA变体，通过可调的融合系数提供灵活的控制每个标准的影响。我们提出了一种系统的参数优化策略，包括核参数选择、子空间维数调整和融合平衡，以及一种替代的核参数优化方法。从小样本（SSS）到大规模设置的图像、表格和信号数据集的广泛实验验证了KLPCDA的适应性。结果表明，在SSS设置下，KLPCDA在识别精度和效率方面始终优于基准方法和卷积神经网络，同时在大规模场景的计算复杂度和存储要求方面保持竞争优势。最后，我们提供了与我们的提案相关的扩展研究主题和未来工作的见解。

{"title":"Kernelized linear principal component discriminant analysis","authors":"Lingxiao Qu , Yan Pei","doi":"10.1016/j.neunet.2026.108539","DOIUrl":"10.1016/j.neunet.2026.108539","url":null,"abstract":"<div><div>In this paper, we propose Kernelized Linear Principal Component Discriminant Analysis (KLPCDA), a structured and unified framework for discriminant analysis that overcomes the fragmentation in existing multi-stage approaches such as PCA+LDA or KPCA+GDA. Instead of treating feature extraction and class discrimination as disjoint steps, KLPCDA formulates a joint optimization model in the Reproducing Kernel Hilbert Space (RKHS), integrating overall variance preservation, between-class separation, and within-class compactness into a fused objective. The formulation supports seven KLPCDA variants, offering flexible control over effects of each criterion through tunable fusion coefficients. We present a systematic parameter optimization strategy, including kernel parameter selection, subspace dimensionality tuning, and fusion balancing, along with an alternative kernel parameter optimization method. Extensive experiments across image, tabular, and signal datasets across small-sample-size (SSS) to larger-scale settings validate the adaptability of KLPCDA. The results demonstrate KLPCDA consistently outperforms benchmark methods and convolutional neural networks in SSS settings on both recognition accuracy and efficiency, while maintaining competitive advantages in computational complexity and storage requirements for large-scale scenarios. Finally, we provide insights into extended study subjects and future work related to our proposal.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"Article 108539"},"PeriodicalIF":6.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145967199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A memristive fuzzy neural network with applications to classification task: A programmable circuit system 记忆模糊神经网络在分类任务中的应用：一个可编程电路系统。

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-07 DOI: 10.1016/j.neunet.2026.108547

Ningye Jiang , Mingxuan Jiang , Jupeng Xie , Haoen Huang , Depeng Li , Zhigang Zeng

Inspired by fuzzy inference systems and neural networks, this paper presents the design of a memristive fuzzy neural network (M-FNN) with applications to classification tasks, implemented in a computing-in-memory (CIM) architecture. Specifically, a modified first-order T-S model is optimized for hardware deployment with a three-dimensional memristor crossbar array (3-D MCA). Based on this model, the overall M-FNN architecture is constructed, where time-controlled schedules enable continuous samples processing. A dedicated writing scheme and multiple functional modules are introduced, along with a peripheral circuit scheme to enhance programmability. Various classification experiments on multiple machine learning datasets demonstrate the adaptability of M-FNN. Error tolerance results further verify its robustness in analog computing. Compared with ASIC-based schemes, the proposed M-FNN achieves higher programmability, allowing flexible adjustment of inputs, rules, and outputs. Additionally, compared with programmable chips such as CPU and FPGA, the proposed M-FNN demonstrates significant improvements in inference speed, integrated area, and power consumption by factors of 1.35 × 10⁵, 914, and 33, respectively.

受模糊推理系统和神经网络的启发，本文提出了一种记忆模糊神经网络（M-FNN）的设计，并将其应用于分类任务，实现在内存计算（CIM）体系结构中。具体来说，改进的一阶T-S模型优化了硬件部署与三维记忆电阻交叉棒阵列（3-D MCA）。基于该模型，构建了整体的M-FNN体系结构，其中时间控制的调度使连续的样本处理成为可能。介绍了一个专用的编写方案和多个功能模块，以及一个外围电路方案，以提高可编程性。在多个机器学习数据集上的分类实验证明了M-FNN的自适应性。容错结果进一步验证了其在模拟计算中的鲁棒性。与基于asic的方案相比，本文提出的M-FNN具有更高的可编程性，可以灵活地调整输入、规则和输出。此外，与CPU和FPGA等可编程芯片相比，所提出的M-FNN在推理速度、集成面积和功耗方面分别提高了1.35 × 105、914和33倍。

{"title":"A memristive fuzzy neural network with applications to classification task: A programmable circuit system","authors":"Ningye Jiang , Mingxuan Jiang , Jupeng Xie , Haoen Huang , Depeng Li , Zhigang Zeng","doi":"10.1016/j.neunet.2026.108547","DOIUrl":"10.1016/j.neunet.2026.108547","url":null,"abstract":"<div><div>Inspired by fuzzy inference systems and neural networks, this paper presents the design of a memristive fuzzy neural network (M-FNN) with applications to classification tasks, implemented in a computing-in-memory (CIM) architecture. Specifically, a modified first-order T-S model is optimized for hardware deployment with a three-dimensional memristor crossbar array (3-D MCA). Based on this model, the overall M-FNN architecture is constructed, where time-controlled schedules enable continuous samples processing. A dedicated writing scheme and multiple functional modules are introduced, along with a peripheral circuit scheme to enhance programmability. Various classification experiments on multiple machine learning datasets demonstrate the adaptability of M-FNN. Error tolerance results further verify its robustness in analog computing. Compared with ASIC-based schemes, the proposed M-FNN achieves higher programmability, allowing flexible adjustment of inputs, rules, and outputs. Additionally, compared with programmable chips such as CPU and FPGA, the proposed M-FNN demonstrates significant improvements in inference speed, integrated area, and power consumption by factors of 1.35 × 10<sup>5</sup>, 914, and 33, respectively.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"197 ","pages":"Article 108547"},"PeriodicalIF":6.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145965954","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Temporal knowledge graphs forecasting based on explainable temporal relation tree-graph 基于可解释时间关系树形图的时间知识图预测

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-07 DOI: 10.1016/j.neunet.2026.108582

Qihong Wu , Ruizhe Ma , Yuan Cheng , Li Yan , Zongmin Ma

In real-world temporal knowledge graphs, relationships among entities often exhibit complex temporal dynamics. Effectively modeling multi-hop temporal relation chains and enabling interpretable reasoning remain core challenges in temporal knowledge graph forecasting, which we address with our proposed model, TRTL (Temporal Relation Tree-based Learning). To tackle these challenges, we introduce a novel reasoning framework grounded in two complementary graph structures: the Sequence Grounding Graph, which captures temporal interval and entity’s relation alignments; the Temporal Relation Tree Graph, which organizes multi-hop relation chains into interpretable and tree-structured reasoning paths. These structures are encoded using a Tree-LSTM enhanced with attention mechanisms, enabling the model to effectively capture temporal logic and long-range dependencies. The tree-based symbolical reasoning process provides interpretable evidence, enhancing the transparency and reliability of predictions. Experiments on two time-interval benchmarks demonstrate that TRTL significantly outperforms existing symbolic-based models.

在现实世界的时间知识图中，实体之间的关系往往表现出复杂的时间动态。有效地建模多跳时间关系链并实现可解释推理仍然是时间知识图预测的核心挑战，我们提出了基于时间关系树的学习模型TRTL （temporal relation Tree-based Learning）。为了解决这些挑战，我们引入了一种基于两个互补图结构的新型推理框架：序列接地图，它捕获时间间隔和实体的关系对齐；时间关系树图，它将多跳关系链组织成可解释的树结构推理路径。这些结构使用带有注意机制的Tree-LSTM进行编码，使模型能够有效地捕获时间逻辑和远程依赖关系。基于树的符号推理过程提供了可解释的证据，提高了预测的透明度和可靠性。在两个时间间隔基准上的实验表明，TRTL显著优于现有的基于符号的模型。

{"title":"Temporal knowledge graphs forecasting based on explainable temporal relation tree-graph","authors":"Qihong Wu , Ruizhe Ma , Yuan Cheng , Li Yan , Zongmin Ma","doi":"10.1016/j.neunet.2026.108582","DOIUrl":"10.1016/j.neunet.2026.108582","url":null,"abstract":"<div><div>In real-world temporal knowledge graphs, relationships among entities often exhibit complex temporal dynamics. Effectively modeling multi-hop temporal relation chains and enabling interpretable reasoning remain core challenges in temporal knowledge graph forecasting, which we address with our proposed model, TRTL (Temporal Relation Tree-based Learning). To tackle these challenges, we introduce a novel reasoning framework grounded in two complementary graph structures: the Sequence Grounding Graph, which captures temporal interval and entity’s relation alignments; the Temporal Relation Tree Graph, which organizes multi-hop relation chains into interpretable and tree-structured reasoning paths. These structures are encoded using a Tree-LSTM enhanced with attention mechanisms, enabling the model to effectively capture temporal logic and long-range dependencies. The tree-based symbolical reasoning process provides interpretable evidence, enhancing the transparency and reliability of predictions. Experiments on two time-interval benchmarks demonstrate that TRTL significantly outperforms existing symbolic-based models.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"Article 108582"},"PeriodicalIF":6.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

NeuroAdaptive multi-resolution integration network for decoding cognitive complexity levels in EEG-based pronoun resolution tasks 基于脑电图的代词识别任务认知复杂性解码的神经自适应多分辨率整合网络。

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-07 DOI: 10.1016/j.neunet.2026.108562

Wenlong Wu , Mengyuan Zhao , Zhong Yin

Pronoun resolution represents a fundamental language comprehension process that varies in cognitive complexity. Prior studies have identified behavioral and neural differences in pronoun processing, but existing models struggle to address background interference of neural activities, capture responses across multiple brain regions and neurophysiological feature domains, and account for subtle differences of complexity-dependent pronoun resolution mechanisms influenced by individual variability. In this study, we investigate three cognitive complexity levels in pronoun resolution including direct-determined, indirect-determined, and undetermined conditions, using electroencephalogram (EEG) recordings from 35 participants. To address these challenges, we developed the NeuroAdaptive Multi-resolution Integration Network (NAMINet) to decode EEG recordings into cognitive complexity conditions. The model employs a spatial encoder to filter background EEG interference by analyzing features across different resolutions and frequency bands. It then uses a dual-domain encoder to capture interactions between statistical and complexity-based EEG features, followed by a fusion module that integrates patterns across frequency, statistical, and complexity domains at each spatial resolution. A context-adaptive training framework ensures robust performance for both individual and cross-participant scenarios. Our results demonstrate distinct neural processing strategies for each complexity level, evidenced by variations in behavioral performance, EEG spectral patterns, and classification accuracies. NAMINet achieved 54.21% (participant-dependent) and 43.59% (cross-participant) classification accuracy, both significantly exceeding chance-level performance. These findings establish the neural substrates of graded cognitive complexity in language processing and provide a computational framework for decoding hierarchical linguistic mechanisms, thus advancing our understanding of brain-language relationships.

代词解析是一个基本的语言理解过程，其认知复杂性各不相同。先前的研究已经确定了代词加工的行为和神经差异，但现有的模型难以解决神经活动的背景干扰，捕捉多个大脑区域和神经生理特征域的反应，并解释受个体差异影响的复杂性依赖代词解决机制的细微差异。在这项研究中，我们利用35名参与者的脑电图（EEG）记录，研究了代词分辨的三种认知复杂性水平，包括直接决定、间接决定和不确定条件。为了解决这些挑战，我们开发了神经自适应多分辨率集成网络（NAMINet），将脑电图记录解码为认知复杂性条件。该模型采用空间编码器，通过分析不同分辨率和频带的特征来过滤背景脑电信号干扰。然后使用双域编码器捕获统计和基于复杂性的EEG特征之间的相互作用，然后使用融合模块在每个空间分辨率下集成跨频率，统计和复杂性域的模式。上下文自适应训练框架可确保个人和跨参与者场景的稳健性能。我们的研究结果表明，在行为表现、脑电图频谱模式和分类准确性方面，每个复杂程度都有不同的神经处理策略。NAMINet实现了54.21%（参与者相关）和43.59%（跨参与者）的分类准确率，均显著超过机会水平的表现。这些发现建立了语言处理中分级认知复杂性的神经基础，并为解码分层语言机制提供了计算框架，从而促进了我们对脑-语言关系的理解。

{"title":"NeuroAdaptive multi-resolution integration network for decoding cognitive complexity levels in EEG-based pronoun resolution tasks","authors":"Wenlong Wu , Mengyuan Zhao , Zhong Yin","doi":"10.1016/j.neunet.2026.108562","DOIUrl":"10.1016/j.neunet.2026.108562","url":null,"abstract":"<div><div>Pronoun resolution represents a fundamental language comprehension process that varies in cognitive complexity. Prior studies have identified behavioral and neural differences in pronoun processing, but existing models struggle to address background interference of neural activities, capture responses across multiple brain regions and neurophysiological feature domains, and account for subtle differences of complexity-dependent pronoun resolution mechanisms influenced by individual variability. In this study, we investigate three cognitive complexity levels in pronoun resolution including direct-determined, indirect-determined, and undetermined conditions, using electroencephalogram (EEG) recordings from 35 participants. To address these challenges, we developed the NeuroAdaptive Multi-resolution Integration Network (NAMINet) to decode EEG recordings into cognitive complexity conditions. The model employs a spatial encoder to filter background EEG interference by analyzing features across different resolutions and frequency bands. It then uses a dual-domain encoder to capture interactions between statistical and complexity-based EEG features, followed by a fusion module that integrates patterns across frequency, statistical, and complexity domains at each spatial resolution. A context-adaptive training framework ensures robust performance for both individual and cross-participant scenarios. Our results demonstrate distinct neural processing strategies for each complexity level, evidenced by variations in behavioral performance, EEG spectral patterns, and classification accuracies. NAMINet achieved 54.21% (participant-dependent) and 43.59% (cross-participant) classification accuracy, both significantly exceeding chance-level performance. These findings establish the neural substrates of graded cognitive complexity in language processing and provide a computational framework for decoding hierarchical linguistic mechanisms, thus advancing our understanding of brain-language relationships.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"197 ","pages":"Article 108562"},"PeriodicalIF":6.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145966907","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A unified framework for EEG seizure detection using universum-integrated generalized eigenvalues proximal support vector machine 基于宇宙积分广义特征值近端支持向量机的脑电图发作检测统一框架

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-07 DOI: 10.1016/j.neunet.2025.108520

Yogesh Kumar, Vrushank Ahire, Mudasir Ganaie

The paper presents novel Universum-enhanced classifiers: the Universum Generalized Eigenvalue Proximal Support Vector Machine (U-GEPSVM) and the Improved U-GEPSVM (IU-GEPSVM) for EEG signal classification. Using the computational efficiency of generalized eigenvalue decomposition and the generalization benefits of Universum learning, the proposed models address critical challenges in EEG analysis: non-stationarity, low signal-to-noise ratio, and limited labeled data. U-GEPSVM extends the GEPSVM framework by incorporating Universum constraints through a ratio-based objective function, while IU-GEPSVM enhances stability through a weighted difference-based formulation that provides independent control over class separation and Universum alignment. The models are evaluated on the Bonn University EEG dataset across two binary classification tasks: (O vs S)-healthy (eyes closed) vs seizure, and (Z vs S)-healthy (eyes open) vs seizure. IU-GEPSVM achieves peak accuracies of 85% (O vs S) and 80% (Z vs S), with mean accuracies of 81.29% and 77.57% respectively, outperforming baseline methods. Rigorous statistical validation confirms these improvements: Friedman tests reveal significant overall differences, pairwise Wilcoxon signed-rank tests with Bonferroni correction establish IU-GEPSVM’s superiority over all baselines, and win-tie-loss analysis demonstrates practical significance. Overall, integrating interictal Universum data yields an efficient and reliable solution for neurological diagnosis.

本文提出了一种新的Universum增强分类器：Universum广义特征值近端支持向量机（U-GEPSVM）和改进的U-GEPSVM （IU-GEPSVM）。利用广义特征值分解的计算效率和Universum学习的泛化优势，所提出的模型解决了脑电分析中的关键挑战：非平稳性、低信噪比和有限的标记数据。U-GEPSVM扩展了GEPSVM框架，通过一个基于比率的目标函数结合Universum的约束，而U-GEPSVM通过一个基于加权差分的公式增强了稳定性，该公式提供了对类分离和Universum校准的独立控制。这些模型在波恩大学EEG数据集上通过两个二元分类任务进行评估：(O vs S)-健康（闭眼）与癫痫发作，(Z vs S)-健康（睁眼）与癫痫发作。u - gepsvm的峰值准确度分别为85% （0 vs S）和80% (Z vs S)，平均准确度分别为81.29%和77.57%，优于基线方法。严格的统计验证证实了这些改进：Friedman检验显示了显著的总体差异，配对Wilcoxon sign -rank检验与Bonferroni校正证实了IU-GEPSVM优于所有基线，而输赢分析显示了实际意义。总的来说，整合Universum的内部数据为神经系统诊断提供了高效可靠的解决方案。

{"title":"A unified framework for EEG seizure detection using universum-integrated generalized eigenvalues proximal support vector machine","authors":"Yogesh Kumar, Vrushank Ahire, Mudasir Ganaie","doi":"10.1016/j.neunet.2025.108520","DOIUrl":"10.1016/j.neunet.2025.108520","url":null,"abstract":"<div><div>The paper presents novel Universum-enhanced classifiers: the Universum Generalized Eigenvalue Proximal Support Vector Machine (U-GEPSVM) and the Improved U-GEPSVM (IU-GEPSVM) for EEG signal classification. Using the computational efficiency of generalized eigenvalue decomposition and the generalization benefits of Universum learning, the proposed models address critical challenges in EEG analysis: non-stationarity, low signal-to-noise ratio, and limited labeled data. U-GEPSVM extends the GEPSVM framework by incorporating Universum constraints through a ratio-based objective function, while IU-GEPSVM enhances stability through a weighted difference-based formulation that provides independent control over class separation and Universum alignment. The models are evaluated on the Bonn University EEG dataset across two binary classification tasks: (O vs S)-healthy (eyes closed) vs seizure, and (Z vs S)-healthy (eyes open) vs seizure. IU-GEPSVM achieves peak accuracies of 85% (O vs S) and 80% (Z vs S), with mean accuracies of 81.29% and 77.57% respectively, outperforming baseline methods. Rigorous statistical validation confirms these improvements: Friedman tests reveal significant overall differences, pairwise Wilcoxon signed-rank tests with Bonferroni correction establish IU-GEPSVM’s superiority over all baselines, and win-tie-loss analysis demonstrates practical significance. Overall, integrating interictal Universum data yields an efficient and reliable solution for neurological diagnosis.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"Article 108520"},"PeriodicalIF":6.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981710","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AMSA-YOLO: Real-time object detection with adaptive multi-scale attention mechanism AMSA-YOLO：基于自适应多尺度注意机制的实时目标检测

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-05 DOI: 10.1016/j.neunet.2026.108545

Canjin Wang , Peng Sun , Chunhui Yang , Xianglong Teng , Rijun Wang

Object detection, as a fundamental task in computer vision, has extensive applications in autonomous driving, video surveillance, medical imaging, and other domains. The YOLO (You Only Look Once) series of algorithms has become the representative method for single-stage object detection due to their excellent real-time performance. However, existing YOLO algorithms still face challenges in small object detection and dense scene detection. This paper proposes AMSA-YOLO (Adaptive Multi-Scale Attention YOLO), an improved YOLO algorithm based on adaptive multi-scale attention mechanism. By introducing scale-aware modules, adaptive spatial attention, and adaptive channel attention, the proposed method significantly improves detection accuracy, particularly for small object detection. Experimental results demonstrate that AMSA-YOLO achieves a 2.3 percentage point improvement in [email protected]:0.95 compared to YOLOv8s on the COCO dataset, with a 3.6 percentage point improvement in small object detection AP, while maintaining inference speed with only a 10.3 % decrease. Significant improvements are also achieved on specialized datasets such as VisDrone and CrowdHuman, proving the effectiveness and practicality of the proposed method.

物体检测作为计算机视觉的一项基础任务，在自动驾驶、视频监控、医学成像等领域有着广泛的应用。YOLO （You Only Look Once）系列算法以其优异的实时性成为单阶段目标检测的代表性方法。然而，现有的YOLO算法在小目标检测和密集场景检测方面仍然面临挑战。本文提出了一种基于自适应多尺度注意机制的改进YOLO算法AMSA-YOLO （Adaptive Multi-Scale Attention YOLO）。通过引入尺度感知模块、自适应空间注意和自适应通道注意，该方法显著提高了检测精度，特别是对小目标的检测。实验结果表明，与COCO数据集上的YOLOv8s相比，AMSA-YOLO在[email protected]:0.95方面提高了2.3个百分点，在小目标检测AP方面提高了3.6个百分点，同时保持了推理速度，仅降低了10.3%。在VisDrone和CrowdHuman等特定数据集上也取得了显著的改进，证明了所提出方法的有效性和实用性。

{"title":"AMSA-YOLO: Real-time object detection with adaptive multi-scale attention mechanism","authors":"Canjin Wang , Peng Sun , Chunhui Yang , Xianglong Teng , Rijun Wang","doi":"10.1016/j.neunet.2026.108545","DOIUrl":"10.1016/j.neunet.2026.108545","url":null,"abstract":"<div><div>Object detection, as a fundamental task in computer vision, has extensive applications in autonomous driving, video surveillance, medical imaging, and other domains. The YOLO (You Only Look Once) series of algorithms has become the representative method for single-stage object detection due to their excellent real-time performance. However, existing YOLO algorithms still face challenges in small object detection and dense scene detection. This paper proposes AMSA-YOLO (Adaptive Multi-Scale Attention YOLO), an improved YOLO algorithm based on adaptive multi-scale attention mechanism. By introducing scale-aware modules, adaptive spatial attention, and adaptive channel attention, the proposed method significantly improves detection accuracy, particularly for small object detection. Experimental results demonstrate that AMSA-YOLO achieves a 2.3 percentage point improvement in [email protected]:0.95 compared to YOLOv8s on the COCO dataset, with a 3.6 percentage point improvement in small object detection AP, while maintaining inference speed with only a 10.3 % decrease. Significant improvements are also achieved on specialized datasets such as VisDrone and CrowdHuman, proving the effectiveness and practicality of the proposed method.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"197 ","pages":"Article 108545"},"PeriodicalIF":6.3,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Leveraging hemispheric asymmetry in structural MRI with an attention-guided 3D CNN for early prediction of Alzheimer’s conversion 利用结构MRI的半球不对称与注意力引导的3D CNN早期预测阿尔茨海默氏症的转换

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-04 DOI: 10.1016/j.neunet.2025.108534

Yi Liu , Lizhong Yao , Jinwei Liu , Haoyu Li , Ying Zeng , Yashu Xu

Early identification of mild cognitive impairment (MCI) progressing to Alzheimer’s disease (AD) is of paramount importance. Despite the notable advances in deep learning in this domain, current approaches are largely based on global brain analysis and often overlook the hemispheric asymmetry, which is a critical biomarker for AD progression. Although longitudinal studies can capture temporal dynamics, their clinical feasibility is constrained by the need for multiple follow-up visits. To address this issue, we propose HemiNet, a lightweight 3D convolutional neural network based on hemispheric difference analysis, enabling accurate prediction of MCI progression from structural MRI at a single time point. HemiNet is designed with three key modules. First, the asymmetry discrepancy mining strategy is employed to quantify interhemispheric structural differences, derive disease-specific biomarkers, and effectively capture multi-level asymmetry features. Second, the contralateral hemispheric fusion mechanism is designed to adaptively unify bilateral features through discrepancy-aware gating combined with depthwise separable convolution, thus strengthening asymmetry patterns indicative of AD. Finally, the pathology focal attention mechanism is applied with sequential channel–spatial attention to highlight pivotal pathological regions, such as the hippocampus and temporal lobe, thereby enhancing the discriminative capacity of the learned features. Extensive experiments and cross-validation on the ADNI dataset demonstrate that HemiNet achieves an AUC of 84.01% and an accuracy of 78.19% for MCI prediction. This study validates the value of hemispheric asymmetry analysis for early AD detection and presents an efficient, lightweight, and interpretable method for MCI progression prediction from a single scan.

早期识别轻度认知障碍（MCI）进展为阿尔茨海默病（AD）是至关重要的。尽管深度学习在这一领域取得了显著进展，但目前的方法主要基于全局大脑分析，往往忽略了半球不对称，而半球不对称是AD进展的关键生物标志物。尽管纵向研究可以捕捉时间动态，但其临床可行性受到多次随访需求的限制。为了解决这个问题，我们提出了HemiNet，这是一个基于半球差异分析的轻量级3D卷积神经网络，能够在单个时间点从结构MRI准确预测MCI进展。HemiNet设计有三个关键模块。首先，采用不对称差异挖掘策略量化半球间结构差异，获得疾病特异性生物标志物，并有效捕获多层次不对称特征。其次，设计对侧半球融合机制，通过差异感知门控结合深度可分离卷积自适应地统一双侧特征，从而加强指示AD的不对称模式。最后，将病理焦点注意机制与顺序通道-空间注意相结合，突出关键病理区域，如海马和颞叶，从而增强对学习特征的辨别能力。在ADNI数据集上进行的大量实验和交叉验证表明，HemiNet预测MCI的AUC为84.01%，准确率为78.19%。本研究验证了半球不对称分析在早期AD检测中的价值，并提出了一种有效、轻量级和可解释的方法，可通过单次扫描预测MCI的进展。

{"title":"Leveraging hemispheric asymmetry in structural MRI with an attention-guided 3D CNN for early prediction of Alzheimer’s conversion","authors":"Yi Liu , Lizhong Yao , Jinwei Liu , Haoyu Li , Ying Zeng , Yashu Xu","doi":"10.1016/j.neunet.2025.108534","DOIUrl":"10.1016/j.neunet.2025.108534","url":null,"abstract":"<div><div>Early identification of mild cognitive impairment (MCI) progressing to Alzheimer’s disease (AD) is of paramount importance. Despite the notable advances in deep learning in this domain, current approaches are largely based on global brain analysis and often overlook the hemispheric asymmetry, which is a critical biomarker for AD progression. Although longitudinal studies can capture temporal dynamics, their clinical feasibility is constrained by the need for multiple follow-up visits. To address this issue, we propose HemiNet, a lightweight 3D convolutional neural network based on hemispheric difference analysis, enabling accurate prediction of MCI progression from structural MRI at a single time point. HemiNet is designed with three key modules. First, the asymmetry discrepancy mining strategy is employed to quantify interhemispheric structural differences, derive disease-specific biomarkers, and effectively capture multi-level asymmetry features. Second, the contralateral hemispheric fusion mechanism is designed to adaptively unify bilateral features through discrepancy-aware gating combined with depthwise separable convolution, thus strengthening asymmetry patterns indicative of AD. Finally, the pathology focal attention mechanism is applied with sequential channel–spatial attention to highlight pivotal pathological regions, such as the hippocampus and temporal lobe, thereby enhancing the discriminative capacity of the learned features. Extensive experiments and cross-validation on the ADNI dataset demonstrate that HemiNet achieves an AUC of 84.01% and an accuracy of 78.19% for MCI prediction. This study validates the value of hemispheric asymmetry analysis for early AD detection and presents an efficient, lightweight, and interpretable method for MCI progression prediction from a single scan.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"197 ","pages":"Article 108534"},"PeriodicalIF":6.3,"publicationDate":"2026-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adaptive dynamic spatial-temporal graph convolutional neural network for traffic flow prediction 交通流预测的自适应动态时空图卷积神经网络

IF 6.3 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2026-01-04 DOI: 10.1016/j.neunet.2025.108529

Yu Jiang , Mingmao Hu , Aihong Gong , Yanfei Lan , Qingshan Gong , Shuaiyu Li , Xu Wang , Zhenghao Yao

Accurate and efficient traffic flow prediction is essential for developing smart cities. Traffic flow data exhibits complex spatio-temporal dependencies, and the weights between nodes may change dynamically due to travel patterns and node attributes. However, existing prediction models primarily rely on static adjacency matrices and complex time series models, which limit model performance. We propose an innovative traffic flow prediction method, the Adaptive Dynamic Spatio-temporal Graph Convolutional Network (ADSTGCN), to address this issue. Specifically, we incorporate a multi-head attention mechanism and an adaptive dynamic adjacency matrix to construct two dynamic spatio-temporal extraction modules, which are integrated with Graph Convolutional Networks (GCN) to overcome the limitation of static adjacency matrices in capturing dynamic spatio-temporal correlations between nodes. At the temporal dimension extraction level, we integrate the Mamba to model long-term time series in traffic flow data, effectively extracting relevant temporal information. Extensive comparative experiments are conducted on four real-world public transportation datasets. The results demonstrate that our model achieves the highest prediction accuracy compared to other baseline models, showcasing its significant potential in traffic flow prediction.

准确、高效的交通流量预测对智慧城市的发展至关重要。交通流数据表现出复杂的时空依赖关系，节点间的权重会因出行模式和节点属性的不同而发生动态变化。然而，现有的预测模型主要依赖于静态邻接矩阵和复杂时间序列模型，这限制了模型的性能。为了解决这一问题，我们提出了一种创新的交通流预测方法——自适应动态时空图卷积网络（ADSTGCN）。具体而言，我们结合多头注意机制和自适应动态邻接矩阵构建了两个动态时空提取模块，并将其与图卷积网络（GCN）相结合，克服了静态邻接矩阵在捕获节点间动态时空相关性方面的局限性。在时间维度提取层面，我们将曼巴模型整合到交通流数据的长期时间序列中，有效提取相关的时间信息。在四个真实的公共交通数据集上进行了广泛的比较实验。结果表明，与其他基线模型相比，我们的模型达到了最高的预测精度，显示了其在交通流预测方面的巨大潜力。

{"title":"Adaptive dynamic spatial-temporal graph convolutional neural network for traffic flow prediction","authors":"Yu Jiang , Mingmao Hu , Aihong Gong , Yanfei Lan , Qingshan Gong , Shuaiyu Li , Xu Wang , Zhenghao Yao","doi":"10.1016/j.neunet.2025.108529","DOIUrl":"10.1016/j.neunet.2025.108529","url":null,"abstract":"<div><div>Accurate and efficient traffic flow prediction is essential for developing smart cities. Traffic flow data exhibits complex spatio-temporal dependencies, and the weights between nodes may change dynamically due to travel patterns and node attributes. However, existing prediction models primarily rely on static adjacency matrices and complex time series models, which limit model performance. We propose an innovative traffic flow prediction method, the Adaptive Dynamic Spatio-temporal Graph Convolutional Network (ADSTGCN), to address this issue. Specifically, we incorporate a multi-head attention mechanism and an adaptive dynamic adjacency matrix to construct two dynamic spatio-temporal extraction modules, which are integrated with Graph Convolutional Networks (GCN) to overcome the limitation of static adjacency matrices in capturing dynamic spatio-temporal correlations between nodes. At the temporal dimension extraction level, we integrate the Mamba to model long-term time series in traffic flow data, effectively extracting relevant temporal information. Extensive comparative experiments are conducted on four real-world public transportation datasets. The results demonstrate that our model achieves the highest prediction accuracy compared to other baseline models, showcasing its significant potential in traffic flow prediction.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"198 ","pages":"Article 108529"},"PeriodicalIF":6.3,"publicationDate":"2026-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145980478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0