NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130727
Wei Meng, Zeyu Huang, Quan Liu, Mincheng Cai, Kun Chen, Li Ma
{"title":"Spike-driven incepformer: A hierarchical spiking transformer with inception-inspired feature learning","authors":"Wei Meng, Zeyu Huang, Quan Liu, Mincheng Cai, Kun Chen, Li Ma","doi":"10.1016/j.neucom.2025.130727","DOIUrl":"10.1016/j.neucom.2025.130727","url":null,"abstract":"<div><div>Designing spike-based attention mechanisms and Transformer architectures has gradually become a hot topic in the field of Spiking Neural Networks (SNNs). Spiking Transformers typically rely on floating-point and integer computations to achieve performance improvements. However, models that maintain Spike-Driven characteristics, while exhibiting lower energy consumption, often suffer from suboptimal performance. This paper proposes an innovative solution to address this trade-off. Firstly, we introduce feature convolution, expanding the receptive field of attention learning through multi-scale feature connections. Secondly, we design a Spike-Driven Feature Attention (SDFA) mechanism, which significantly reduces computational complexity and enhances performance by utilizing feature matrix operations. Thirdly, we integrate the Inception structure into the Spike-Driven Transformer, replacing traditional MLP layers. Finally, we incorporate diverse branch convolutions to mitigate information loss caused by neurons in the final layer. Experimental results demonstrate that the Spike-Driven Incepformer achieves excellent performance while balancing parameter count and computational cost. On the ImageNet-1k dataset, it attains an accuracy of 80.41%, representing the state-of-the-art for Spike-Driven SNNs. These findings provide new insights for designing low-energy, high-performance spiking neural networks and promote the application of spiking Transformers in broader artificial intelligence domains. Code will be available at <span><span>https://github.com/2ephyrus/SDIncepformer</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130727"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144365116","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130649
Yongfeng Bu , Haoxiang Liang , Huansheng Song , Shijie Sun , Zhaoyang Zhang
{"title":"Causally-guided graph Mamba for detecting socially abnormal vehicle trajectories","authors":"Yongfeng Bu , Haoxiang Liang , Huansheng Song , Shijie Sun , Zhaoyang Zhang","doi":"10.1016/j.neucom.2025.130649","DOIUrl":"10.1016/j.neucom.2025.130649","url":null,"abstract":"<div><div>With the rapid development of autonomous driving and internet of things (IoT) technologies, vehicle trajectory anomaly detection becomes one of the important tasks in road monitoring to help traffic data collection and intelligent traffic management. Most methods use simple vehicle interaction with static information to accomplish this task. In contrast, long sequences of trajectory spatio-temporal information can better describe the state of vehicle traveling. In this study, we introduce a trajectory anomaly detection pipeline for spatio-temporal graph modeling to address the challenges in this task. First, we propose a causally guided dynamic graph representation (CGDG) to efficiently model key interactions in complex trajectories. This bootstrapping and decoupling mechanism lays the foundation for our subsequent process. Subsequently, we parse the dynamic evolution process in the spatiotemporal graph by using spatio-tempora feature fusion Mamba (STFMamba) as an encoder and decoder. The anomaly results are generated by reconstruction probabilities. To address the challenge of scarce and expensive datasets in trajectory anomaly detection, we propose the TRAREAL benchmark dataset supplemented with various anomalous event scenarios for experiments. Our method performs well in the evaluation of natural trajectory benchmark datasets. The source codes are available at <span><span>https://github.com/yongfengB/DATMamba</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"649 ","pages":"Article 130649"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144470587","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":"Oscillations enhance time-series prediction in reservoir computing with feedback","authors":"Yuji Kawai , Takashi Morita , Jihoon Park , Minoru Asada","doi":"10.1016/j.neucom.2025.130728","DOIUrl":"10.1016/j.neucom.2025.130728","url":null,"abstract":"<div><div>Reservoir computing, a machine learning framework used for modeling the brain, can predict temporal data with little observations and minimal computational resources. However, it is difficult to accurately reproduce the long-term target time series because the reservoir system becomes unstable. This predictive capability is required for a wide variety of time-series processing, including predictions of motor timing and chaotic dynamical systems. This study proposes oscillation-driven reservoir computing (ODRC) with feedback, where oscillatory signals are fed into a reservoir network to stabilize the network activity and induce complex reservoir dynamics. The ODRC can reproduce long-term target time series more accurately than conventional reservoir computing methods in a motor timing and chaotic time-series prediction tasks. Furthermore, it generates a time series similar to the target in the unexperienced period, that is, it can learn the abstract generative rules from limited observations. Given these significant improvements made by the simple and computationally inexpensive implementation, the ODRC would serve as a practical model of various time series data. Moreover, we will discuss biological implications of the ODRC, considering it as a model of neural oscillations and their cerebellar processors.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130728"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144331271","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130705
Yixiao Hu , Haolin Wang , Jiaxiang Cao , Baobin Li
{"title":"Dataset-adaptive and bias-constrained brain age estimation using pyramid squeeze and excitation transformer","authors":"Yixiao Hu , Haolin Wang , Jiaxiang Cao , Baobin Li","doi":"10.1016/j.neucom.2025.130705","DOIUrl":"10.1016/j.neucom.2025.130705","url":null,"abstract":"<div><div>Modeling the biological changes of the human brain is crucial for identifying brain-related diseases and health monitoring. The brain age predicted from MRI data is one useful biomarker for quantifying the maturation and ageing process of human brain. However, the acquisition and preprocessing of MRI data can introduce significant variations between datasets, making it essential to develop models with higher accuracy and robustness for cross-dataset evaluation. To achieve this goal, our paper combines the strengths of CNNs and transformers, proposing the Pyramid Squeeze and Excitation Transformer (PSET) as a novel approach for brain age estimation. In the PSET framework, 3D inception blocks function as an advanced CNN module to capture localized features while the self-attention mechanism is integrated with a squeeze-and-excitation module to extract global features across disparate patches. In particular, a dataset-adaptive and bias-constrained (DABC) model training strategy is proposed to improve the robustness for cross-dataset situations and reduce the bias by introducing self-supervised pre-training, meta-learning and novel loss functions. Experiment results on the dataset of 15,437 healthy brain T1-MRIs (MAE=2.342), demonstrated that the proposed method outperforms both classic visual models and existing brain age estimation models, in the aspect of accuracy, generality and unbiasedness. Additionally, through visualization analysis, we identified the key brain regions that play significant roles in brain age estimation, including the occipital lobe. We compared the brain age gap between patients with diseases and healthy control groups, demonstrating the phenomenon of abnormal aging in conditions such as Alzheimer’s disease and mild cognitive impairment.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130705"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322434","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130706
Feng Liu, Yixin Huang, Yin Li, Qiuheng Wang
{"title":"CrossFingerprint: A multi-level fusion network for 3D fingerprint verification","authors":"Feng Liu, Yixin Huang, Yin Li, Qiuheng Wang","doi":"10.1016/j.neucom.2025.130706","DOIUrl":"10.1016/j.neucom.2025.130706","url":null,"abstract":"<div><div>Contactless 3D fingerprint gains significant attention in recent years by considering hygiene and template safety. Many previous studies on 3D fingerprint verification relied heavily on minutiae-based approaches, necessitating intricate extracting procedures. To alleviate these problems, this paper presents a two-stage multi-fusion network for fingerprint verification called CrossFingerprint. In the pretraining phase, it achieves a 2D-3D correspondence of fingerprints by maximizing the agreement between point clouds and their corresponding rendered 2D images in the invariant space. In the fine-tuning phase, two weight-shared encoders are designed to directly extract features from 3D fingerprints. Within this framework, the fusion of 2D images and 3D point clouds occurs at both the feature and score levels, resulting in a further improvement in matching accuracy. The experiments on a public database and an in-house database verify that our approach achieves 1.21% in Equal Error Rate (EER) and significantly speeds up feature extraction and matching compared with state-of-the-art approaches based on minutiae.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130706"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144335694","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130713
Tingting Liu , Yujue Cai , Guiping Chen , Hongguang Wei , Junqi Bai , Yuan Liu , Xiubao Sui , Qian Chen
{"title":"Adversarial network for unsupervised infrared image colorization based on full-scale feature fusion and cosine contrastive learning","authors":"Tingting Liu , Yujue Cai , Guiping Chen , Hongguang Wei , Junqi Bai , Yuan Liu , Xiubao Sui , Qian Chen","doi":"10.1016/j.neucom.2025.130713","DOIUrl":"10.1016/j.neucom.2025.130713","url":null,"abstract":"<div><div>Thermal infrared images, unaffected by lighting and haze, are widely used in security surveillance, autonomous vehicles, and nighttime traffic monitoring. However, their grayscale nature lacks color and texture details, limiting applications in image recognition and object detection. Converting infrared images to daytime color enhances visual perception and broadens their utility. Despite advancements in infrared image colorization, challenges such as texture distortion, detail blurring, and poor image quality persist. To address these issues, a novel unsupervised learning framework, termed Cosine Contrastive Learning Generative Adversarial Network (CCLGAN), is proposed. Firstly, the traditional UNet architecture is improved by introducing full-scale skip connections and deep supervision. Full-scale skip connections integrate low-level details with high-level semantic features, while deep supervision aids in learning hierarchical feature maps. Additionally, a parameter-free neuron-based 3D attention mechanism is incorporated into the Mamba module to capture long-range dependencies and enable effective feature selection and fusion. Secondly, a novel contrastive loss function is designed, incorporating cosine distance metrics into the traditional contrastive loss framework. By maximizing cosine decision margins and normalizing, intra-class variance is minimized, and inter-class variance is maximized, ensuring consistency between input infrared image patches and output color image patches. Finally, extensive comparative analysis on common datasets demonstrates that the proposed method outperforms existing state-of-the-art techniques in colorization performance. This research advances infrared image processing and enhances the visual quality of converted images. The code is available at <span><span>https://github.com/LTTdouble/CCLGAN</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"649 ","pages":"Article 130713"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144481274","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130790
Jinjun Zhou , Si Zhou , Xinyi Wang , Junneng Shao , Qing Lu
{"title":"Meta-guided dual path convolutional neural network for depression diagnosis with functional MR images","authors":"Jinjun Zhou , Si Zhou , Xinyi Wang , Junneng Shao , Qing Lu","doi":"10.1016/j.neucom.2025.130790","DOIUrl":"10.1016/j.neucom.2025.130790","url":null,"abstract":"<div><div>Deep learning methods with functional magnetic resonance imaging (fMRI) data are successful in diagnosis of depression. However, developing robust models remains challenging due to small datasets and individual heterogeneity. Domain knowledge has the potential to enhance deep-learning-based diagnosis. Previous imaging studies reported abnormalities in brain regions, and meta-analysis can identify spatially convergent abnormal regions. In the present study, we proposed a meta-guided deep-learning framework integrating meta-analysis findings as domain knowledge. We designed the preprocessing of the meta-map for framework integration and developed meta convolutional block and meta dual-path block using the meta-map’s disease-associated regions as spatial guidance to learn ReHo features. Our framework achieved 76.9 % accuracy using a cohort of 385 subjects (192 healthy controls and 193 depressed patients). The effectiveness of the meta-map was comprehensively validated through extensive comparative experiments and systematic ablation studies. Experiments validated the meta-map’s effectiveness in addressing limited sample size and heterogeneity issues. This study introduces the findings of traditional literature beyond the given medical dataset, providing a more promising approach to addressing the problem of small-sized medical datasets for psychiatric disorders.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"649 ","pages":"Article 130790"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502548","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130719
Paraskevi Chasani, Aristidis Likas
{"title":"Statistical modeling of univariate multimodal data","authors":"Paraskevi Chasani, Aristidis Likas","doi":"10.1016/j.neucom.2025.130719","DOIUrl":"10.1016/j.neucom.2025.130719","url":null,"abstract":"<div><div>Unimodality constitutes a key property indicating grouping behavior of the data around a single mode of its density. We propose a method that partitions univariate data into unimodal subsets through recursive splitting around valley points of the data density. For valley point detection, we introduce properties of critical points on the convex hull of the empirical cumulative density function (ecdf) plot that provide indications on the existence of density valleys. Next, we apply a unimodal data modeling approach that provides a statistical model for each obtained unimodal subset in the form of a Uniform Mixture Model (UMM). Consequently, a hierarchical statistical model of the initial dataset is obtained in the form of a mixture of UMMs, named as the Unimodal Mixture Model (UDMM). The proposed method is non-parametric, hyperparameter-free, automatically estimates the number of unimodal subsets and provides accurate statistical models as indicated by experimental results on clustering and density estimation tasks.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130719"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144331273","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}
NeurocomputingPub Date : 2025-06-18DOI: 10.1016/j.neucom.2025.130667
Yuting Feng, Tao Yang, Kaidi Wang, Jiali Sun, Yushu Yu
{"title":"Variable admittance control via Reinforcement Learning: Enhancing UAV interactions across diverse platforms","authors":"Yuting Feng, Tao Yang, Kaidi Wang, Jiali Sun, Yushu Yu","doi":"10.1016/j.neucom.2025.130667","DOIUrl":"10.1016/j.neucom.2025.130667","url":null,"abstract":"<div><div>A compliant control model based on Reinforcement Learning (RL) is proposed to allow UAVs (Unmanned Aerial Vehicles) to interact with the environment more effectively and autonomously execute force control tasks. The model learns an optimal admittance adjustment policy for interaction and simultaneously optimizes energy consumption and trajectory tracking of the UAV state. This facilitates stable manipulation of UAVs in unknown environments with interaction forces. Furthermore, the model ensures safe, compliant, and flexible interaction while safeguarding the UAV’s external structures from damage. To assess the model performance, we validated the approach in a simulation environment using a UAV. The model was also tested across different UAV types and various low-level control parameters, demonstrating superior performance in all scenarios. Additionally, we applied this methodology to two distinct UAV types used in real-world applications. Empirical evidence shows that our proposed methods consistently achieve superior results. We also applied similar methodologies to verify 6D interaction in a simulation of a fully actuated platform consisting of three UAVs. Using a high-level training strategy, we evaluated the platform’s ability to slide along a bevel and achieve optimal results in our comparative experiments.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130667"},"PeriodicalIF":5.5,"publicationDate":"2025-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144335697","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}
NeurocomputingPub Date : 2025-06-17DOI: 10.1016/j.neucom.2025.130648
Tong Ouyang, Bo Ma, Hao Xu
{"title":"Similar concept matters: Prototype analogy for few-shot classification","authors":"Tong Ouyang, Bo Ma, Hao Xu","doi":"10.1016/j.neucom.2025.130648","DOIUrl":"10.1016/j.neucom.2025.130648","url":null,"abstract":"<div><div>Few-shot classification aims to produce a classifier to recognize novel classes not seen during training with few labeled samples. The unseen classes and scarce samples make few-shot classification truly challenging. Transfer learning has been demonstrated to be an efficient paradigm for few-shot classification problem in recent literature. However, most methods based on transfer learning only utilize the parameters of the pre-trained model, ignoring the base features themselves which can be seen as the semantic concepts learned by the pre-trained model. In this paper, our main innovation lies in highlighting the importance of the learned semantic concepts of base classes. We propose a simple yet effective approach for few-shot classification to explore the pre-trained semantic features of base classes. Our approach innovatively employs prototype analogy inside and outside the few-shot classification task, to perform clustering and to select base features respectively, in an alternate and iterative way. We further design the best arrangement for these two steps. The initial centroids for clustering are constantly optimized by more and more accurate base features which are selected by the clustered novel prototypes from the previous iteration. When the iteration converges, the best semantic base features are selected to complete the prototypes of novel classes. Extensive experiments on four standard datasets and two deep backbones are conducted to demonstrate the effectiveness of our proposed prototype analogy method. Notably, our method requires neither sophisticated transductive algorithm nor additional learnable parameters besides the pre-trained model yet achieving comparable or even state-of-the-art performance on the miniImageNet, tieredImageNet and CIFAR-FS datasets.</div></div>","PeriodicalId":19268,"journal":{"name":"Neurocomputing","volume":"648 ","pages":"Article 130648"},"PeriodicalIF":5.5,"publicationDate":"2025-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144313571","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}