Neural Networks : the Official Journal of the International Neural Network Society

Reinforcement learning has achieved promising results on robotic control tasks but struggles to leverage information effectively from multiple sensory modalities that differ in many characteristics. Recent works construct auxiliary losses based on reconstruction or mutual information to extract joint representations from multiple sensory inputs to improve the sample efficiency and performance of reinforcement learning algorithms. However, the representations learned by these methods could capture information irrelevant to learning a policy and may degrade the performance...

The complex and diverse practical background drives this paper to explore a new neurodynamic approach (NA) to solve nonsmooth interval-valued optimization problems (IVOPs) constrained by interval partial order and more general sets. On the one hand, to deal with the uncertainty of interval-valued information, the LU-optimality condition of IVOPs is established through a deterministic form. On the other hand, according to the penalty method and adaptive controller, the interval partial order constraint and set constraint are punished by one adaptive parameter, which is a key enabler for the feasibility of states while having a lower solution space dimension and avoiding estimating exact penalty parameters...

Electroencephalography (EEG) based Brain Computer Interface (BCI) systems play a significant role in facilitating how individuals with neurological impairments effectively interact with their environment. In real world applications of BCI system for clinical assistance and rehabilitation training, the EEG classifier often needs to learn on sequentially arriving subjects in an online manner. As patterns of EEG signals can be significantly different for different subjects, the EEG classifier can easily erase knowledge of learnt subjects after learning on later ones as it performs decoding in online streaming scenario, namely catastrophic forgetting...

The great learning ability of deep learning facilitates us to comprehend the real physical world, making learning to simulate complicated particle systems a promising endeavour both in academia and industry. However, the complex laws of the physical world pose significant challenges to the learning based simulations, such as the varying spatial dependencies between interacting particles and varying temporal dependencies between particle system states in different time stamps, which dominate particles' interacting behavior and the physical systems' evolution patterns...

Reinforcement Learning (RL) is a significant machine learning subfield that emphasizes learning actions based on environment to obtain optimal behavior policy. RL agents can make decisions at variable time scales in the form of temporal abstractions, also known as options. The issue of discovering options has seen a considerable research effort. Most notably, the Interest Option Critic (IOC) algorithm first extends the initial set to the interest function, providing a method for learning options specialized to certain state space regions...

In order to enhance the performance of Transformer models for long-term multivariate forecasting while minimizing computational demands, this paper introduces the Joint Time-Frequency Domain Transformer (JTFT). JTFT combines time and frequency domain representations to make predictions. The frequency domain representation efficiently extracts multi-scale dependencies while maintaining sparsity by utilizing a small number of learnable frequencies. Simultaneously, the time domain (TD) representation is derived from a fixed number of the most recent data points, strengthening the modeling of local relationships and mitigating the effects of non-stationarity...

In this work, we demonstrate the training, conversion, and implementation flow of an FPGA-based bin-ratio ensemble spiking neural network applied for radioisotope identification. The combination of techniques including learned step quantisation (LSQ) and pruning facilitated the implementation by compressing the network's parameters down to 30% yet retaining the accuracy of 97.04% with an accuracy loss of less than 1%. Meanwhile, the proposed ensemble network of 20 3-layer spiking neural networks (SNNs), which incorporates 1160 spiking neurons, only needs 334 μs for a single inference with the given clock frequency of 100 MHz...

In this paper, we show that the Kolmogorov two hidden layer neural network model with a continuous, discontinuous bounded and unbounded activation function in the second hidden layer can precisely represent continuous, discontinuous bounded and all unbounded multivariate functions, respectively.

Given a graph G, the network collapse problem (NCP) selects a vertex subset S of minimum cardinality from G such that the difference in the values of a given measure function f(G)-f(G∖S) is greater than a predefined collapse threshold. Many graph analytic applications can be formulated as NCPs with different measure functions, which often pose a significant challenge due to their NP-hard nature. As a result, traditional greedy algorithms, which select the vertex with the highest reward at each step, may not effectively find the optimal solution...

Spiking neural networks (SNNs), as the brain-inspired neural networks, encode information in spatio-temporal dynamics. They have the potential to serve as low-power alternatives to artificial neural networks (ANNs) due to their sparse and event-driven nature. However, existing SNN-based models for pixel-level semantic segmentation tasks suffer from poor performance and high memory overhead, failing to fully exploit the computational effectiveness and efficiency of SNNs. To address these challenges, we propose the multi-scale and full spike segmentation network (MFS-Seg), which is based on the deep direct trained SNN and represents the first attempt to train a deep SNN with surrogate gradients for semantic segmentation...

This paper investigates the dynamics of a directed acyclic neural network by edge adding control. We find that the local stability and Hopf bifurcation of the controlled network only depend on the size and intersection of directed cycles, instead of the number and position of the added edges. More specifically, if there is no cycle in the controlled network, the local dynamics of the network will remain unchanged and Hopf bifurcation will not occur even if the number of added edges is sufficient. However, if there exist cycles, then the network may undergo Hopf bifurcation...

Few-shot Event Detection (FSED) aims to identify novel event types in new domains with very limited annotated data. Previous PN-based (Prototypical Network) joint methods suffer from insufficient learning of token-wise label dependency and inaccurate prototypes. To solve these problems, we propose a span-based FSED model, called SpanFSED, which decomposes FSED into two subprocesses, including span extractor and event classifier. In span extraction, we convert sequential labels into a global boundary matrix that enables the span extractor to acquire precise boundary information irrespective of label dependency...

Recent temporal action detection models have focused on end-to-end trainable approaches to utilize the representational power of backbone networks. Despite the advantages of end-to-end trainable methods, these models still employ a small spatial resolution (e.g., 96 × 96) due to the inefficient trade-off between computational cost and spatial resolution. In this study, we argue that a simple pooling method (e.g., adaptive average pooling) acts as a bottleneck at the spatial aggregation part, restricting representational power...

Few-shot video action recognition (FS-VAR) is a challenging task that requires models to have significant expressive power in order to identify previously unseen classes using only a few labeled examples. However, due to the limited number of support samples, the model's performance is highly sensitive to the distribution of the sampled data. The representativeness of the support data is insufficient to cover the entire class, and the support features may contain shared information that confuses the classifier, leading to biased classification...

In recent years, distributed stochastic algorithms have become increasingly useful in the field of machine learning. However, similar to traditional stochastic algorithms, they face a challenge where achieving high fitness on the training set does not necessarily result in good performance on the test set. To address this issue, we propose to use of a distributed network topology to improve the generalization ability of the algorithms. We specifically focus on the Sharpness-Aware Minimization (SAM) algorithm, which relies on perturbation weights to find the maximum point with better generalization ability...

In the realm of long document classification (LDC), previous research has predominantly focused on modeling unimodal texts, overlooking the potential of multi-modal documents incorporating images. To address this gap, we introduce an innovative approach for multi-modal long document classification based on the Hierarchical Prompt and Multi-modal Transformer (HPMT). The proposed HPMT method facilitates multi-modal interactions at both the section and sentence levels, enabling a comprehensive capture of hierarchical structural features and complex multi-modal associations of long documents...

The rhythm of bonafide speech is often difficult to replicate, which causes that the fundamental frequency (F0) of synthetic speech is significantly different from that of real speech. It is expected that the F0 feature contains the discriminative information for the fake speech detection (FSD) task. In this paper, we propose a novel F0 subband for FSD. In addition, to effectively model the F0 subband so as to improve the performance of FSD, the spatial reconstructed local attention Res2Net (SR-LA Res2Net) is proposed...

Generalized zero-shot learning (GZSL) aims to recognize both seen and unseen classes, while only samples from seen classes are available for training. The mainstream methods mitigate the lack of unseen training data by simulating the visual unseen samples. However, the sample generator is actually learned with just seen-class samples, and semantic descriptions of unseen classes are just provided to the pre-trained sample generator for unseen data generation, therefore, the generator would have bias towards seen categories, and the unseen generation quality, including both precision and diversity, is still the main learning challenge...

In this article, we introduce a system of non-linear inverse mixed variational inequalities (SNIMVIs). We propose a proximal neurodynamic model (PNDM) for solving SNIMVIs, leveraging proximal mappings. The uniqueness of the continuous solution for the PNDM is proved by assuming Lipschitz continuity. Moreover, we establish the global asymptotic stability of equilibrium points of the PNDM, contingent upon Lipschitz continuity and strong monotonicity. Additionally, an iterative algorithm involving proximal mappings for solving the SNIMVIs is presented...

We propose a new score-based model with one-step sampling. Previously, score-based models were burdened with heavy computations due to iterative sampling. For substituting the iterative process, we train a standalone generator to compress all the time steps with the gradient backpropagated from the score network. In order to produce meaningful gradients for the generator, the score network is trained to simultaneously match the real data distribution and mismatch the fake data distribution. This model has the following advantages: (1) For sampling, it generates a fake image with only one step forward...