International journal of neural systems最新文献

This research addresses the interplay of changing social and organizational context factors with the big five personality traits and the three main characterizing elements of burnout. A computational analysis is contributed based on an integrative biological-mental-social network modeling approach. The simulation results show how two people who are high in personality traits such as agreeableness, openness, extraversion, conscientiousness, and highly sensitive to neuroticism, are vulnerable to reaching a burnout level in all dimensions whenever the organizational context is changing in a less favorable direction. By a What-If analysis, it is analyzed how important characteristics affect the outcomes and indicate how, in a qualitative sense, that is in line with empirical literature. Several differentiations are made. In particular, the connection between the three dimensions of burnout shows that it is possible that one employee reaches a burnout state while the other does not. It is also shown how therapy alone may not be sufficient as a long-term treatment, but therapy of one employee does affect the other. As numerical data are not (yet) available, further numerical validation has been proposed for future work.

This paper introduces a comprehensive framework for clustering analysis, centered on a novel incremental silhouette score calculation designed specifically for hierarchical clustering. This innovative method significantly reduces the computational complexity of silhouette evaluation, transforming the process from O(K N) to effectively O(N) for K hierarchical configurations (demonstrated by an over 100-fold speedup in our tests) making it feasible for large-scale datasets and enabling efficient cluster number estimation within hierarchical clustering scenarios. Building on this, we revisit and enhance the Principal Direction Divisive Partitioning (IPDDP) algorithm, proposing principal component analysis-maximum margin divisive clustering (PCA-MMDC), which utilizes multiple principal components for more accurate data partitioning, and PCA-MMDC-sc, which incorporates a scatter-based cluster selection for improved balance. These are integrated into a hybrid clustering strategy that combines the strengths of incremental silhouette calculation and the enhanced algorithms, allowing for robust cluster identification and effective management of noise and outliers. Experimental results on synthetic and real-world datasets demonstrate notable improvements in clustering accuracy (achieving an average Adjusted Rand Index (ARI) increase of over 10 percentage points on custom noisy synthetic datasets compared to K-Means) and computational efficiency. While the choice of principal components in PCA-MMDC presents a parameter, the overall framework offers a scalable and robust solution for complex clustering tasks, with future work aimed at adaptive parameter selection and extending incremental calculations to other validation metrics.

Machine learning, deep learning and neural networks are extensively developed in many fields, with neural networks playing an important role in a wide variety of applications. However, a sufficient explanation of the structure and functionality of complex and deep neural networks is still needed. In this paper, it is shown that bio-inspired networks are useful for the explanation of network functions. First, the asymmetric network is created based on the biological retinal networks. Second, the classification performance of the asymmetric network is compared to that of the symmetric networks. The directional vectors in the asymmetric networks are generated on the adjacent neurons caused by movement stimulus, which create independent subspaces. Vectors for the movement stimulus are reported experimentally to be generated in the layered cortex in the brain. In this paper, it is shown computationally that many directional movement vectors are generated in the layered asymmetric networks, which create also independent subspaces. Further, when the correlational activities of the adjacent cells are represented in the directed vectors, they create independent subspaces than the direct inputs in the networks. These asymmetric subnetworks will facilitate the transmission of sensory information to higher-level processes such as efficient feature extraction, classification, and learning in the layered networks.

Virus machines, which develop models of computation inspired by biological processes and the spread of viruses among hosts, deviate from the traditional methods. These virus machines are recognized for their computational power (functioning as algorithms) and their ability to tackle computationally difficult problems. In this paper, we introduce a new extension of the matrix-based representation of virus machines. In this way, hosts, the number of viruses and the instructions to control virus transmission are represented as vectors and matrices, describing the computations of virus machines by linear algebra operations. We also use our matrix representation to show invariants, useful in the proofs, of such machines. In addition, an explicit example is shown to clarify the computation and invariants using the representation. That is, a virus machine that computes the discrete logarithm, which relies on the presumed intractability of cryptosystems such the digital signature algorithm.

In this paper, we propose a spiking neural network model with Hebbian connectivity for implementing energy-efficient associative memory, whose activity is determined by input stimuli. The model consists of three interacting layers of Hodgkin-Huxley-Mainen spiking neurons with excitatory and inhibitory synaptic connections. Information patterns are stored in memory using a symmetric Hebbian matrix and can be retrieved in response to a specific stimulus pattern. Binary images are encoded using in-phase and anti-phase oscillations relative to a global clock signal. Utilizing the phase-locking effect allows for cluster synchronization of neurons (both on the input and output layers). Interneurons in the intermediate layer filter signal propagation pathways depending on the context of the input layer, effectively engaging only a portion of the synaptic connections within the Hebbian matrix for recognition. The stability of the oscillation phase is investigated for both in-phase and anti-phase synchronization modes when recognizing direct and inverse images. This context-dependent effect opens promising avenues for the development of analog hardware circuits for energy-efficient neurocomputing applications, potentially leading to breakthroughs in artificial intelligence and cognitive computing.

Spiking Neural Networks (SNNs), as a biologically inspired computational model, have garnered significant attention in object detection and image classification due to their event-driven mechanism and low-power characteristics. However, in object detection tasks, the residual structures in conventional networks introduce nonspiking operations, posing a critical challenge for SNNs. To address this issue, we propose a multi-layer feature cascade fusion SNN (MFCF-SNN) for object detection. During feature extraction, our novel multi-level cascaded feature extraction module replaces residual connections with cascade operations, eliminating nonspiking computations while enhancing gradient propagation to deeper layers. For downsampling, we introduce a pooling-convolution module that combines max-pooling and spiking convolution, effectively preserving feature information and improving gradient flow. These two modules collectively ensure pure spike-based computation while facilitating deep network training, thereby enhancing detection accuracy. Experimental results on the PASCAL VOC 2012 and SSDD datasets demonstrate state-of-the-art performance, validating the effectiveness of our approach in advancing SNN-based object detection.

Spiking Neural P Systems are parallel and distributed computational models inspired by biological neurons, emerging from membrane computing and applied to solving computationally difficult problems. This paper focuses on the computational complexity of such systems using neuron division rules and colored spikes for the SAT problem. We prove a conjecture stated in a recent paper, showing that enhancing the model with an input module reduces computing time. Additionally, we prove that the inclusion of budding rules extends the model's capability to solve all problems in the complexity class PSPACE. These findings advance research on Spiking Neural P Systems and their application to complex problems; however, whether both budding rules and division rules are required to extend these methods to problem domains beyond the NP class remains an open question.

Although a variety of deep learning-based methods have been introduced for Salient Object Detection (SOD) to RGB and Depth (RGB-D) images, existing approaches still encounter challenges, including inadequate cross-modal feature fusion, significant errors in saliency estimation due to noise in depth information, and limited model generalization capabilities. To tackle these challenges, this paper introduces an innovative method for RGB-D SOD, TranSNP-Net, which integrates Nonlinear Spiking Neural P (NSNP) systems with Transformer networks. TranSNP-Net effectively fuses RGB and depth features by introducing an enhanced feature fusion module (SNPFusion) and an attention mechanism. Unlike traditional methods, TranSNP-Net leverages fine-tuned Swin (shifted window transformer) as its backbone network, significantly improving the model's generalization performance. Furthermore, the proposed hierarchical feature decoder (SNP-D) notably enhances accuracy in complex scenes where depth noise is prevalent. According to the experimental findings, the mean scores for the four metrics S-measure, F-measure, E-measure and MEA on the six RGB-D benchmark datasets are 0.9328, 0.9356, 0.9558 and 0.0288. TranSNP-Net achieves superior performance compared to 14 leading methods in six RGB-D benchmark datasets.

Thermal and RGB images exhibit significant differences in information representation, especially in low-light or nighttime environments. Thermal images provide temperature information, complementing the RGB images by restoring details and contextual information. However, the spatial discrepancy between different modalities in RGB-Thermal (RGB-T) semantic segmentation tasks complicates the process of multimodal feature fusion, leading to a loss of spatial contextual information and limited model performance. This paper proposes a channel-space fusion nonlinear spiking neural P system model network (CSPM-SNPNet) to address these challenges. This paper designs a novel color-thermal image fusion module to effectively integrate features from both modalities. During decoding, a nonlinear spiking neural P system is introduced to enhance multi-channel information extraction through the convolution of spiking neural P systems (ConvSNP) operations, fully restoring features learned in the encoder. Experimental results on public datasets MFNet and PST900 demonstrate that CSPM-SNPNet significantly improves segmentation performance. Compared with the existing methods, CSPM-SNPNet achieves a 0.5% improvement in mIOU on MFNet and 1.8% on PST900, showcasing its effectiveness in complex scenes.

The models used to predict epileptic seizures based on electroencephalogram (EEG) signals often encounter substantial challenges due to the requirement for large, labeled datasets and the inherent complexity of EEG data, which hinders their robustness and generalization capability. This study proposes CLResNet, a framework for predicting epileptic seizures, which combines contrastive self-supervised learning with a modified deep residual neural network to address the above challenges. In contrast to traditional models, CLResNet uses unlabeled EEG data for pre-training to extract robust feature representations. It is then fine-tuned on a smaller labeled dataset to significantly reduce its reliance on labeled data while improving its efficiency and predictive accuracy. The contrastive learning (CL) framework enhances the ability of the model to distinguish between preictal and interictal states, thus improving its robustness and generalizability. The architecture of CLResNet contains residual connections that enable it to learn deep features of the data and ensure an efficient gradient flow. The results of the evaluation of the model on the CHB-MIT dataset showed that it outperformed prevalent methods in the field, with an accuracy of 92.97%, sensitivity of 94.18%, and false-positive rate of 0.043/h. On the Siena dataset, the model also achieved competitive performance, with an accuracy of 92.79%, a sensitivity of 91.47%, and a false-positive rate of 0.041/h. These results confirm the effectiveness of CLResNet in addressing variations in EEG data, and show that contrastive self-supervised learning is a robust and accurate approach for predicting seizures.