Neural Networks最新文献_第4页

A unified noise and watermark removal from information bottleneck-based modeling 从基于信息瓶颈的建模中统一去除噪声和水印。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-04 DOI: 10.1016/j.neunet.2024.106853

Hanjuan Huang , Hsing-Kuo Pao

Both image denoising and watermark removal aim to restore a clean image from an observed noisy or watermarked one. The past research consists of the non-learning type with limited effectiveness or the learning types with limited interpretability. To address these issues simultaneously, we propose a method to deal with both the image-denoising and watermark removal tasks in a unified approach. The noises and watermarks are both considered to have different nuisance patterns from the original image content, therefore should be detected by robust image analysis. The unified detection method is based on the well-known information bottleneck (IB) theory and the proposed SIB-GAN where image content and nuisance patterns are well separated by a supervised approach. The IB theory guides us to keep the valuable content such as the original image by a controlled compression on the input (the noisy or watermark-included image) and then only the content without the nuisances can go through the network for effective noise or watermark removal. Additionally, we adjust the compression parameter in IB theory to learn a representation that approaches the minimal sufficient representation of the image content. In particular, to deal with the non-blind noises, an appropriate amount of compression can be estimated from the solid theory foundation. Working on the denoising task given the unseen data with blind noises also shows the model’s generalization power. All of the above shows the interpretability of the proposed method. Overall, the proposed method has achieved promising results across three tasks: image denoising, watermark removal, and mixed noise and watermark removal, obtaining resultant images very close to the original image content and owning superior performance to almost all state-of-the-art approaches that deal with the same tasks.

图像去噪和去除水印的目的都是从观测到的噪声或水印图像中还原出干净的图像。过去的研究包括效果有限的非学习型研究或可解释性有限的学习型研究。为了同时解决这些问题，我们提出了一种统一处理图像去噪和去水印任务的方法。噪声和水印都被认为具有不同于原始图像内容的干扰模式，因此应通过鲁棒图像分析来检测。统一检测方法是基于著名的信息瓶颈（IB）理论和所提出的 SIB-GAN 方法，即通过监督方法将图像内容和干扰模式很好地分离开来。IB 理论指导我们通过对输入（有噪声或包含水印的图像）进行有控制的压缩来保留有价值的内容，如原始图像，然后只有没有干扰的内容才能通过网络进行有效的噪声或水印去除。此外，我们还会调整 IB 理论中的压缩参数，以学习一种接近图像内容最小充分表示的表示方法。特别是在处理非盲目噪声时，可以从坚实的理论基础中估算出适当的压缩量。在处理未见过的盲噪声数据时，去噪任务也显示了模型的泛化能力。所有这些都表明了所提出方法的可解释性。总之，所提出的方法在图像去噪、去除水印、去除混合噪声和水印这三个任务中都取得了可喜的成果，得到的图像非常接近原始图像内容，其性能几乎优于处理相同任务的所有先进方法。

{"title":"A unified noise and watermark removal from information bottleneck-based modeling","authors":"Hanjuan Huang , Hsing-Kuo Pao","doi":"10.1016/j.neunet.2024.106853","DOIUrl":"10.1016/j.neunet.2024.106853","url":null,"abstract":"<div><div>Both image denoising and watermark removal aim to restore a clean image from an observed noisy or watermarked one. The past research consists of the non-learning type with limited effectiveness or the learning types with limited interpretability. To address these issues simultaneously, we propose a method to deal with both the image-denoising and watermark removal tasks in a unified approach. The noises and watermarks are both considered to have different nuisance patterns from the original image content, therefore should be detected by robust image analysis. The unified detection method is based on the well-known information bottleneck (IB) theory and the proposed SIB-GAN where image content and nuisance patterns are well separated by a supervised approach. The IB theory guides us to keep the valuable content such as the original image by a controlled compression on the input (the noisy or watermark-included image) and then only the content without the nuisances can go through the network for effective noise or watermark removal. Additionally, we adjust the compression parameter in IB theory to learn a representation that approaches the minimal sufficient representation of the image content. In particular, to deal with the non-blind noises, an appropriate amount of compression can be estimated from the solid theory foundation. Working on the denoising task given the unseen data with blind noises also shows the model’s generalization power. All of the above shows the interpretability of the proposed method. Overall, the proposed method has achieved promising results across three tasks: image denoising, watermark removal, and mixed noise and watermark removal, obtaining resultant images very close to the original image content and owning superior performance to almost all state-of-the-art approaches that deal with the same tasks.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106853"},"PeriodicalIF":6.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142644984","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

FedART: A neural model integrating federated learning and adaptive resonance theory FedART：融合了联合学习和自适应共振理论的神经模型。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-04 DOI: 10.1016/j.neunet.2024.106845

Shubham Pateria, Budhitama Subagdja, Ah-Hwee Tan

Federated Learning (FL) has emerged as a promising paradigm for collaborative model training across distributed clients while preserving data privacy. However, prevailing FL approaches aggregate the clients’ local models into a global model through multi-round iterative parameter averaging. This leads to the undesirable bias of the aggregated model towards certain clients in the presence of heterogeneous data distributions among the clients. Moreover, such approaches are restricted to supervised classification tasks and do not support unsupervised clustering. To address these limitations, we propose a novel one-shot FL approach called Federated Adaptive Resonance Theory (FedART) which leverages self-organizing Adaptive Resonance Theory (ART) models to learn category codes, where each code represents a cluster of similar data samples. In FedART, the clients learn to associate their private data with various local category codes. Under heterogeneity, the local codes across different clients represent heterogeneous data. In turn, a global model takes these local codes as inputs and aggregates them into global category codes, wherein heterogeneous client data is indirectly represented by distinctly encoded global codes, in contrast to the averaging out of parameters in the existing approaches. This enables the learned global model to handle heterogeneous data. In addition, FedART employs a universal learning mechanism to support both federated classification and clustering tasks. Our experiments conducted on various federated classification and clustering tasks show that FedART consistently outperforms state-of-the-art FL methods on data with heterogeneous distribution across clients.

联盟学习（FL）已成为跨分布式客户端协同模型训练的一种有前途的模式，同时还能保护数据隐私。然而，现有的联合学习方法通过多轮迭代参数平均将客户机的本地模型聚合成一个全局模型。这就导致了在客户端之间存在异构数据分布的情况下，聚合模型会对某些客户端产生不良偏差。此外，这种方法仅限于监督分类任务，不支持无监督聚类。为了解决这些局限性，我们提出了一种名为 "联合自适应共振理论（FedART）"的新颖的单次 FL 方法，它利用自组织自适应共振理论（ART）模型来学习类别代码，其中每个代码代表一个相似数据样本集群。在 FedART 中，客户端学习将其私人数据与各种本地类别代码相关联。在异质性条件下，不同客户端的本地代码代表异质性数据。反过来，全局模型将这些本地代码作为输入，并将其聚合为全局类别代码，其中，异构客户端数据由不同编码的全局代码间接表示，这与现有方法中的参数平均化形成鲜明对比。这使得学习到的全局模型能够处理异构数据。此外，FedART 还采用了一种通用学习机制，以支持联盟分类和聚类任务。我们在各种联合分类和聚类任务上进行的实验表明，在客户端异构分布的数据上，FedART 的表现始终优于最先进的 FL 方法。

{"title":"FedART: A neural model integrating federated learning and adaptive resonance theory","authors":"Shubham Pateria, Budhitama Subagdja, Ah-Hwee Tan","doi":"10.1016/j.neunet.2024.106845","DOIUrl":"10.1016/j.neunet.2024.106845","url":null,"abstract":"<div><div>Federated Learning (FL) has emerged as a promising paradigm for collaborative model training across distributed clients while preserving data privacy. However, prevailing FL approaches aggregate the clients’ local models into a global model through multi-round iterative parameter averaging. This leads to the undesirable bias of the aggregated model towards certain clients in the presence of heterogeneous data distributions among the clients. Moreover, such approaches are restricted to supervised classification tasks and do not support unsupervised clustering. To address these limitations, we propose a novel one-shot FL approach called Federated Adaptive Resonance Theory (FedART) which leverages self-organizing Adaptive Resonance Theory (ART) models to learn category codes, where each code represents a cluster of similar data samples. In FedART, the clients learn to associate their private data with various local category codes. Under heterogeneity, the local codes across different clients represent heterogeneous data. In turn, a global model takes these local codes as inputs and aggregates them into global category codes, wherein heterogeneous client data is indirectly represented by distinctly encoded global codes, in contrast to the averaging out of parameters in the existing approaches. This enables the learned global model to handle heterogeneous data. In addition, FedART employs a universal learning mechanism to support both federated classification and clustering tasks. Our experiments conducted on various federated classification and clustering tasks show that FedART consistently outperforms state-of-the-art FL methods on data with heterogeneous distribution across clients.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106845"},"PeriodicalIF":6.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142631299","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

Global practical finite-time synchronization of disturbed inertial neural networks by delayed impulsive control 通过延迟脉冲控制实现受干扰惯性神经网络的全球实用有限时间同步

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-04 DOI: 10.1016/j.neunet.2024.106873

Qian Cui , Jinde Cao , Mahmoud Abdel-Aty , Ardak Kashkynbayev

This paper delves into the practical finite-time synchronization (FTS) problem for inertial neural networks (INNs) with external disturbances. Firstly, based on Lyapunov theory, the local practical FTS of INNs with bounded external disturbances can be realized by effective finite time control. Then, building upon the local results, we extend the synchronization to a global practical level under delayed impulsive control. By designing appropriate hybrid controllers, the global practical FTS criteria of disturbed INNs are obtained and the corresponding settling time is estimated. In addition, for impulsive control, the maximum impulsive interval is used to describe the frequency at which the impulses occur. We optimize the maximum impulsive interval, aiming to minimize impulses occurrence, which directly translates to reduced control costs. Moreover, by comparing the global FTS results for INNs without external disturbances, it can be found that the existence of perturbations necessitates either higher impulsive intensity or denser impulses to maintain networks synchronization. Two examples are shown to demonstrate the reasonableness of designed hybrid controllers.

本文深入研究了具有外部干扰的惯性神经网络（INN）的实用有限时间同步（FTS）问题。首先，基于李亚普诺夫理论，通过有效的有限时间控制，可以实现有界外部干扰的 INNs 的局部实用 FTS。然后，在局部结果的基础上，我们将同步扩展到延迟脉冲控制下的全局实用水平。通过设计适当的混合控制器，我们得到了受扰 INN 的全局实用有限时间控制标准，并估算了相应的稳定时间。此外，对于脉冲控制，最大脉冲间隔用于描述脉冲发生的频率。我们对最大脉冲间隔进行优化，旨在将脉冲发生率降至最低，从而直接降低控制成本。此外，通过比较无外部扰动 INN 的全局 FTS 结果，可以发现扰动的存在需要更高的脉冲强度或更密集的脉冲来保持网络同步。两个实例证明了所设计的混合控制器的合理性。

{"title":"Global practical finite-time synchronization of disturbed inertial neural networks by delayed impulsive control","authors":"Qian Cui , Jinde Cao , Mahmoud Abdel-Aty , Ardak Kashkynbayev","doi":"10.1016/j.neunet.2024.106873","DOIUrl":"10.1016/j.neunet.2024.106873","url":null,"abstract":"<div><div>This paper delves into the practical finite-time synchronization (FTS) problem for inertial neural networks (INNs) with external disturbances. Firstly, based on Lyapunov theory, the local practical FTS of INNs with bounded external disturbances can be realized by effective finite time control. Then, building upon the local results, we extend the synchronization to a global practical level under delayed impulsive control. By designing appropriate hybrid controllers, the global practical FTS criteria of disturbed INNs are obtained and the corresponding settling time is estimated. In addition, for impulsive control, the maximum impulsive interval is used to describe the frequency at which the impulses occur. We optimize the maximum impulsive interval, aiming to minimize impulses occurrence, which directly translates to reduced control costs. Moreover, by comparing the global FTS results for INNs without external disturbances, it can be found that the existence of perturbations necessitates either higher impulsive intensity or denser impulses to maintain networks synchronization. Two examples are shown to demonstrate the reasonableness of designed hybrid controllers.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106873"},"PeriodicalIF":6.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142631304","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

HirMTL: Hierarchical Multi-Task Learning for dense scene understanding HirMTL：用于密集场景理解的分层多任务学习。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-04 DOI: 10.1016/j.neunet.2024.106854

Huilan Luo , Weixia Hu , Yixiao Wei , Jianlong He , Minghao Yu

In the realm of artificial intelligence, simultaneous multi-task learning is crucial, particularly for dense scene understanding. To address this, we introduce HirMTL, a novel hierarchical multi-task learning framework designed to enhance dense scene analysis. HirMTL is adept at facilitating interaction at the scale level, ensuring task-adaptive multi-scale feature fusion, and fostering task-level feature interchange. It leverages the inherent correlations between tasks to create a synergistic learning environment. Initially, HirMTL enables concurrent sharing and fine-tuning of features at the single-scale level. This is further extended by the Task-Adaptive Fusion module (TAF), which intelligently blends features across scales, specifically attuned to each task’s unique requirements. Complementing this, the Asymmetric Information Comparison Module (AICM) skillfully differentiates and processes both shared and unique features, significantly refining task-specific performance with enhanced accuracy. Our extensive experiments on various dense prediction tasks validate HirMTL’s exceptional capabilities, showcasing its superiority over existing multi-task learning models and underscoring the benefits of its hierarchical approach.

在人工智能领域，同步多任务学习至关重要，尤其是在理解密集场景时。为此，我们推出了 HirMTL，这是一种新颖的分层多任务学习框架，旨在加强密集场景分析。HirMTL 擅长促进尺度级别的交互，确保任务自适应的多尺度特征融合，并促进任务级别的特征交换。它利用任务之间固有的相关性创造了一个协同学习环境。最初，HirMTL 可在单尺度级别实现特征的并发共享和微调。任务自适应融合模块（TAF）进一步扩展了这一功能，它能智能地融合不同尺度的特征，特别适合每个任务的独特要求。作为补充，非对称信息比较模块（AICM）能巧妙地区分和处理共享特征和独特特征，从而显著提高特定任务的性能和准确性。我们在各种密集预测任务上进行了大量实验，验证了 HirMTL 的卓越能力，展示了它优于现有多任务学习模型的优势，并强调了其分层方法的好处。

{"title":"HirMTL: Hierarchical Multi-Task Learning for dense scene understanding","authors":"Huilan Luo , Weixia Hu , Yixiao Wei , Jianlong He , Minghao Yu","doi":"10.1016/j.neunet.2024.106854","DOIUrl":"10.1016/j.neunet.2024.106854","url":null,"abstract":"<div><div>In the realm of artificial intelligence, simultaneous multi-task learning is crucial, particularly for dense scene understanding. To address this, we introduce HirMTL, a novel hierarchical multi-task learning framework designed to enhance dense scene analysis. HirMTL is adept at facilitating interaction at the scale level, ensuring task-adaptive multi-scale feature fusion, and fostering task-level feature interchange. It leverages the inherent correlations between tasks to create a synergistic learning environment. Initially, HirMTL enables concurrent sharing and fine-tuning of features at the single-scale level. This is further extended by the Task-Adaptive Fusion module (TAF), which intelligently blends features across scales, specifically attuned to each task’s unique requirements. Complementing this, the Asymmetric Information Comparison Module (AICM) skillfully differentiates and processes both shared and unique features, significantly refining task-specific performance with enhanced accuracy. Our extensive experiments on various dense prediction tasks validate HirMTL’s exceptional capabilities, showcasing its superiority over existing multi-task learning models and underscoring the benefits of its hierarchical approach.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106854"},"PeriodicalIF":6.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142631413","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 protocol for trustworthy EEG decoding with neural networks 利用神经网络进行可信脑电图解码的协议。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-02 DOI: 10.1016/j.neunet.2024.106847

Davide Borra , Elisa Magosso , Mirco Ravanelli

Deep learning solutions have rapidly emerged for EEG decoding, achieving state-of-the-art performance on a variety of decoding tasks. Despite their high performance, existing solutions do not fully address the challenge posed by the introduction of many hyperparameters, defining data pre-processing, network architecture, network training, and data augmentation. Automatic hyperparameter search is rarely performed and limited to network-related hyperparameters. Moreover, pipelines are highly sensitive to performance fluctuations due to random initialization, hindering their reliability. Here, we design a comprehensive protocol for EEG decoding that explores the hyperparameters characterizing the entire pipeline and that includes multi-seed initialization for providing robust performance estimates. Our protocol is validated on 9 datasets about motor imagery, P300, SSVEP, including 204 participants and 26 recording sessions, and on different deep learning models. We accompany our protocol with extensive experiments on the main aspects influencing it, such as the number of participants used for hyperparameter search, the split into sequential simpler searches (multi-step search), the use of informed vs. non-informed search algorithms, and the number of random seeds for obtaining stable performance. The best protocol included 2-step hyperparameter search via an informed search algorithm, with the final training and evaluation performed using 10 random initializations. The optimal trade-off between performance and computational time was achieved by using a subset of 3–5 participants for hyperparameter search. Our protocol consistently outperformed baseline state-of-the-art pipelines, widely across datasets and models, and could represent a standard approach for neuroscientists for decoding EEG in a trustworthy and reliable way.

用于脑电图解码的深度学习解决方案迅速崛起，在各种解码任务中取得了最先进的性能。尽管性能很高，但现有解决方案并不能完全解决引入许多超参数所带来的挑战，这些超参数定义了数据预处理、网络架构、网络训练和数据扩增。自动超参数搜索很少执行，而且仅限于与网络相关的超参数。此外，管道对随机初始化导致的性能波动非常敏感，从而影响了其可靠性。在此，我们设计了一种用于脑电图解码的综合协议，该协议可探索表征整个管道的超参数，并包括多种子初始化，以提供稳健的性能估计。我们的方案在有关运动图像、P300、SSVEP 的 9 个数据集（包括 204 名参与者和 26 个记录会话）以及不同的深度学习模型上进行了验证。我们对影响协议的主要方面进行了大量实验，如用于超参数搜索的参与者人数、将搜索分为连续的简单搜索（多步骤搜索）、使用知情搜索算法与非知情搜索算法，以及获得稳定性能的随机种子数量。最佳方案包括通过知情搜索算法进行两步超参数搜索，最后使用 10 个随机初始化进行训练和评估。通过使用 3-5 个参与者子集进行超参数搜索，实现了性能和计算时间之间的最佳权衡。我们的方案在广泛的数据集和模型中始终优于最先进的基线管道，可以代表神经科学家以可信和可靠的方式解码脑电图的标准方法。

{"title":"A protocol for trustworthy EEG decoding with neural networks","authors":"Davide Borra , Elisa Magosso , Mirco Ravanelli","doi":"10.1016/j.neunet.2024.106847","DOIUrl":"10.1016/j.neunet.2024.106847","url":null,"abstract":"<div><div>Deep learning solutions have rapidly emerged for EEG decoding, achieving state-of-the-art performance on a variety of decoding tasks. Despite their high performance, existing solutions do not fully address the challenge posed by the introduction of many hyperparameters, defining data pre-processing, network architecture, network training, and data augmentation. Automatic hyperparameter search is rarely performed and limited to network-related hyperparameters. Moreover, pipelines are highly sensitive to performance fluctuations due to random initialization, hindering their reliability. Here, we design a comprehensive protocol for EEG decoding that explores the hyperparameters characterizing the entire pipeline and that includes multi-seed initialization for providing robust performance estimates. Our protocol is validated on 9 datasets about motor imagery, P300, SSVEP, including 204 participants and 26 recording sessions, and on different deep learning models. We accompany our protocol with extensive experiments on the main aspects influencing it, such as the number of participants used for hyperparameter search, the split into sequential simpler searches (multi-step search), the use of informed vs. non-informed search algorithms, and the number of random seeds for obtaining stable performance. The best protocol included 2-step hyperparameter search via an informed search algorithm, with the final training and evaluation performed using 10 random initializations. The optimal trade-off between performance and computational time was achieved by using a subset of 3–5 participants for hyperparameter search. Our protocol consistently outperformed baseline state-of-the-art pipelines, widely across datasets and models, and could represent a standard approach for neuroscientists for decoding EEG in a trustworthy and reliable way.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"182 ","pages":"Article 106847"},"PeriodicalIF":6.0,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142645080","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Skill matters: Dynamic skill learning for multi-agent cooperative reinforcement learning 技能很重要：多代理合作强化学习的动态技能学习

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-02 DOI: 10.1016/j.neunet.2024.106852

Tong Li , Chenjia Bai , Kang Xu , Chen Chu , Peican Zhu , Zhen Wang

With the popularization of intelligence, the necessity of cooperation between intelligent machines makes the research of collaborative multi-agent reinforcement learning (MARL) more extensive. Existing approaches typically address this challenge through task decomposition of the environment or role classification of agents. However, these studies may rely on the sharing of parameters between agents, resulting in the homogeneity of agent behavior, which is not effective for complex tasks. Or training that relies on external rewards is difficult to adapt to scenarios with sparse rewards. Based on the above challenges, in this paper we propose a novel dynamic skill learning (DSL) framework for agents to learn more diverse abilities motivated by internal rewards. Specifically, the DSL has two components: (i) Dynamic skill discovery, which encourages the production of meaningful skills by exploring the environment in an unsupervised manner, using the inner product between a skill vector and a trajectory representation to generate intrinsic rewards. Meanwhile, the Lipschitz constraint of the state representation function is used to ensure the proper trajectory of the learned skills. (ii) Dynamic skill assignment, which utilizes a policy controller to assign skills to each agent based on its different trajectory latent variables. In addition, in order to avoid training instability caused by frequent changes in skill selection, we introduce a regularization term to limit skill switching between adjacent time steps. We thoroughly tested the DSL approach on two challenging benchmarks, StarCraft II and Google Research Football. Experimental results show that compared with strong benchmarks such as QMIX and RODE, DSL effectively improves performance and is more adaptable to difficult collaborative scenarios.

随着智能的普及，智能机器之间合作的必要性使得协作式多代理强化学习（MARL）的研究更加广泛。现有方法通常通过对环境进行任务分解或对代理进行角色分类来应对这一挑战。然而，这些研究可能依赖于代理之间的参数共享，导致代理行为的同质性，这对于复杂任务来说并不有效。或者依赖外部奖励的训练难以适应奖励稀少的场景。基于上述挑战，我们在本文中提出了一种新颖的动态技能学习（DSL）框架，让代理在内部奖励的激励下学习更多样化的能力。具体来说，动态技能学习有两个组成部分：(i) 动态技能发现，它鼓励以无监督的方式探索环境，利用技能向量和轨迹表示之间的内积产生内在奖励，从而产生有意义的技能。同时，利用状态表示函数的 Lipschitz 约束来确保所学技能的正确轨迹。(ii) 动态技能分配，即利用策略控制器，根据每个代理的不同轨迹潜变量为其分配技能。此外，为了避免技能选择的频繁变化导致训练不稳定，我们引入了正则化项来限制相邻时间步之间的技能切换。我们在《星际争霸 II》和谷歌研究足球这两个具有挑战性的基准上对 DSL 方法进行了全面测试。实验结果表明，与 QMIX 和 RODE 等强基准相比，DSL 有效地提高了性能，并且更能适应困难的协作场景。

{"title":"Skill matters: Dynamic skill learning for multi-agent cooperative reinforcement learning","authors":"Tong Li , Chenjia Bai , Kang Xu , Chen Chu , Peican Zhu , Zhen Wang","doi":"10.1016/j.neunet.2024.106852","DOIUrl":"10.1016/j.neunet.2024.106852","url":null,"abstract":"<div><div>With the popularization of intelligence, the necessity of cooperation between intelligent machines makes the research of collaborative multi-agent reinforcement learning (MARL) more extensive. Existing approaches typically address this challenge through task decomposition of the environment or role classification of agents. However, these studies may rely on the sharing of parameters between agents, resulting in the homogeneity of agent behavior, which is not effective for complex tasks. Or training that relies on external rewards is difficult to adapt to scenarios with sparse rewards. Based on the above challenges, in this paper we propose a novel dynamic skill learning (DSL) framework for agents to learn more diverse abilities motivated by internal rewards. Specifically, the DSL has two components: (i) Dynamic skill discovery, which encourages the production of meaningful skills by exploring the environment in an unsupervised manner, using the inner product between a skill vector and a trajectory representation to generate intrinsic rewards. Meanwhile, the Lipschitz constraint of the state representation function is used to ensure the proper trajectory of the learned skills. (ii) Dynamic skill assignment, which utilizes a policy controller to assign skills to each agent based on its different trajectory latent variables. In addition, in order to avoid training instability caused by frequent changes in skill selection, we introduce a regularization term to limit skill switching between adjacent time steps. We thoroughly tested the DSL approach on two challenging benchmarks, StarCraft II and Google Research Football. Experimental results show that compared with strong benchmarks such as QMIX and RODE, DSL effectively improves performance and is more adaptable to difficult collaborative scenarios.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106852"},"PeriodicalIF":6.0,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142631417","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

Deep Incomplete Multi-view Clustering via Multi-level Imputation and Contrastive Alignment 通过多级估算和对比对齐进行深度不完整多视角聚类

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-02 DOI: 10.1016/j.neunet.2024.106851

Ziyu Wang, Yiming Du, Yao Wang, Rui Ning, Lusi Li

Deep incomplete multi-view clustering (DIMVC) aims to enhance clustering performance by capturing consistent information from incomplete multiple views using deep models. Most existing DIMVC methods typically employ imputation-based strategies to handle missing views before clustering. However, they often assume complete data availability across all views, overlook potential low-quality views, and perform imputation at a single data level, leading to challenges in accurately inferring missing data. To address these issues, we propose a novel imputation-based approach called Multi-level Imputation and Contrastive Alignment (MICA) to simultaneously improve imputation quality and boost clustering performance. Specifically, MICA employs an individual deep model for each view, which unifies view feature learning and cluster assignment prediction. It leverages the learned features from available instances to construct an adaptive cross-view graph for reliable view selection. Guided by these reliable views, MICA performs multi-level (feature-level, data-level, and reconstruction-level) imputation to preserve topological structures across levels and ensure accurate missing feature inference. The complete features are then used for discriminative cluster assignment learning. Additionally, an instance- and cluster-level contrastive alignment is conducted on the cluster assignments to further enhance semantic consistency across views. Experimental results show the effectiveness and superior performance of the proposed MICA method.

深度不完整多视图聚类（DIMVC）旨在利用深度模型捕捉不完整多视图中的一致信息，从而提高聚类性能。现有的大多数 DIMVC 方法通常采用基于估算的策略，在聚类前处理缺失视图。然而，这些方法通常假定所有视图中的数据都是完整可用的，忽略了潜在的低质量视图，并且只在单一数据级别上执行归因，从而导致在准确推断缺失数据方面面临挑战。为了解决这些问题，我们提出了一种新颖的基于估算的方法，称为多层次估算和对比对齐（MICA），以同时提高估算质量和聚类性能。具体来说，MICA 为每个视图采用一个单独的深度模型，将视图特征学习和聚类分配预测统一起来。它利用从可用实例中学习到的特征来构建自适应跨视图图，从而实现可靠的视图选择。在这些可靠视图的指导下，MICA 执行多级（特征级、数据级和重构级）归因，以保留跨级拓扑结构，确保准确的缺失特征推断。然后，完整的特征将用于判别性聚类分配学习。此外，还对聚类分配进行了实例和聚类级别的对比对齐，以进一步增强跨视图的语义一致性。实验结果表明了所提出的 MICA 方法的有效性和优越性能。

{"title":"Deep Incomplete Multi-view Clustering via Multi-level Imputation and Contrastive Alignment","authors":"Ziyu Wang, Yiming Du, Yao Wang, Rui Ning, Lusi Li","doi":"10.1016/j.neunet.2024.106851","DOIUrl":"10.1016/j.neunet.2024.106851","url":null,"abstract":"<div><div>Deep incomplete multi-view clustering (DIMVC) aims to enhance clustering performance by capturing consistent information from incomplete multiple views using deep models. Most existing DIMVC methods typically employ imputation-based strategies to handle missing views before clustering. However, they often assume complete data availability across all views, overlook potential low-quality views, and perform imputation at a single data level, leading to challenges in accurately inferring missing data. To address these issues, we propose a novel imputation-based approach called Multi-level Imputation and Contrastive Alignment (MICA) to simultaneously improve imputation quality and boost clustering performance. Specifically, MICA employs an individual deep model for each view, which unifies view feature learning and cluster assignment prediction. It leverages the learned features from available instances to construct an adaptive cross-view graph for reliable view selection. Guided by these reliable views, MICA performs multi-level (feature-level, data-level, and reconstruction-level) imputation to preserve topological structures across levels and ensure accurate missing feature inference. The complete features are then used for discriminative cluster assignment learning. Additionally, an instance- and cluster-level contrastive alignment is conducted on the cluster assignments to further enhance semantic consistency across views. Experimental results show the effectiveness and superior performance of the proposed MICA method.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106851"},"PeriodicalIF":6.0,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142607185","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

Tensorial multiview low-rank high-order graph learning for context-enhanced domain adaptation 用于情境增强领域适应的张量多视图低阶高阶图学习。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-02 DOI: 10.1016/j.neunet.2024.106859

Chenyang Zhu , Lanlan Zhang , Weibin Luo , Guangqi Jiang , Qian Wang

Unsupervised Domain Adaptation (UDA) is a machine learning technique that facilitates knowledge transfer from a labeled source domain to an unlabeled target domain, addressing distributional discrepancies between these domains. Existing UDA methods often fail to effectively capture and utilize contextual relationships within the target domain. This research introduces a novel framework called Tensorial Multiview Low-Rank High-Order Graph Learning (MLRGL), which addresses these challenges by learning high-order graphs constrained by low-rank tensors to uncover contextual relations. The proposed framework ensures prediction consistency between randomly masked target images and their pseudo-labels by leveraging spatial context to generate multiview domain-invariant features through various augmented masking techniques. A high-order graph is constructed by combining Laplacian graphs to propagate these multiview features. Low-rank constraints are applied along both horizontal and vertical dimensions to better uncover inter-view and inter-class correlations among multiview features. This high-order graph is used to create an affinity matrix, mapping multiview features into a unified subspace. Prototype vectors and unsupervised clustering are then employed to calculate conditional probabilities for UDA tasks. We evaluated our approach using three different backbones across three benchmark datasets. The results demonstrate that the MLRGL framework outperforms current state-of-the-art methods in various UDA tasks. Additionally, our framework exhibits robustness to hyperparameter variations and demonstrates that multiview approaches outperform single-view solutions.

无监督领域适应（UDA）是一种机器学习技术，它有助于将知识从有标签的源领域转移到无标签的目标领域，解决这些领域之间的分布差异问题。现有的 UDA 方法往往不能有效捕捉和利用目标域中的上下文关系。这项研究引入了一个名为张量多视图低阶高阶图学习（MLRGL）的新框架，通过学习受低阶张量约束的高阶图来揭示上下文关系，从而应对这些挑战。所提出的框架利用空间上下文，通过各种增强遮挡技术生成多视图域不变特征，从而确保随机遮挡的目标图像与其伪标签之间的预测一致性。通过结合拉普拉斯图构建高阶图来传播这些多视图特征。为了更好地揭示多视图特征之间的视图间和类间相关性，在水平和垂直维度上都应用了低秩约束。这种高阶图用于创建亲和矩阵，将多视角特征映射到统一的子空间中。然后利用原型向量和无监督聚类来计算 UDA 任务的条件概率。我们在三个基准数据集上使用三种不同的骨干对我们的方法进行了评估。结果表明，在各种 UDA 任务中，MLRGL 框架优于当前最先进的方法。此外，我们的框架对超参数变化具有鲁棒性，并证明多视角方法优于单视角解决方案。

{"title":"Tensorial multiview low-rank high-order graph learning for context-enhanced domain adaptation","authors":"Chenyang Zhu , Lanlan Zhang , Weibin Luo , Guangqi Jiang , Qian Wang","doi":"10.1016/j.neunet.2024.106859","DOIUrl":"10.1016/j.neunet.2024.106859","url":null,"abstract":"<div><div>Unsupervised Domain Adaptation (UDA) is a machine learning technique that facilitates knowledge transfer from a labeled source domain to an unlabeled target domain, addressing distributional discrepancies between these domains. Existing UDA methods often fail to effectively capture and utilize contextual relationships within the target domain. This research introduces a novel framework called Tensorial Multiview Low-Rank High-Order Graph Learning (MLRGL), which addresses these challenges by learning high-order graphs constrained by low-rank tensors to uncover contextual relations. The proposed framework ensures prediction consistency between randomly masked target images and their pseudo-labels by leveraging spatial context to generate multiview domain-invariant features through various augmented masking techniques. A high-order graph is constructed by combining Laplacian graphs to propagate these multiview features. Low-rank constraints are applied along both horizontal and vertical dimensions to better uncover inter-view and inter-class correlations among multiview features. This high-order graph is used to create an affinity matrix, mapping multiview features into a unified subspace. Prototype vectors and unsupervised clustering are then employed to calculate conditional probabilities for UDA tasks. We evaluated our approach using three different backbones across three benchmark datasets. The results demonstrate that the MLRGL framework outperforms current state-of-the-art methods in various UDA tasks. Additionally, our framework exhibits robustness to hyperparameter variations and demonstrates that multiview approaches outperform single-view solutions.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106859"},"PeriodicalIF":6.0,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142607221","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

Contrastive fine-grained domain adaptation network for EEG-based vigilance estimation. 基于脑电图的警觉性估计的对比性细粒度域适应网络。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-01 Epub Date: 2024-08-08 DOI: 10.1016/j.neunet.2024.106617

Kangning Wang, Wei Wei, Weibo Yi, Shuang Qiu, Huiguang He, Minpeng Xu, Dong Ming

Vigilance state is crucial for the effective performance of users in brain-computer interface (BCI) systems. Most vigilance estimation methods rely on a large amount of labeled data to train a satisfactory model for the specific subject, which limits the practical application of the methods. This study aimed to build a reliable vigilance estimation method using a small amount of unlabeled calibration data. We conducted a vigilance experiment in the designed BCI-based cursor-control task. Electroencephalogram (EEG) signals of eighteen participants were recorded in two sessions on two different days. And, we proposed a contrastive fine-grained domain adaptation network (CFGDAN) for vigilance estimation. Here, an adaptive graph convolution network (GCN) was built to project the EEG data of different domains into a common space. The fine-grained feature alignment mechanism was designed to weight and align the feature distributions across domains at the EEG channel level, and the contrastive information preservation module was developed to preserve the useful target-specific information during the feature alignment. The experimental results show that the proposed CFGDAN outperforms the compared methods in our BCI vigilance dataset and SEED-VIG dataset. Moreover, the visualization results demonstrate the efficacy of the designed feature alignment mechanisms. These results indicate the effectiveness of our method for vigilance estimation. Our study is helpful for reducing calibration efforts and promoting the practical application potential of vigilance estimation methods.

警觉状态对于脑机接口（BCI）系统中用户的有效表现至关重要。大多数警觉性估计方法都依赖于大量标记数据来为特定对象训练一个令人满意的模型，这限制了这些方法的实际应用。本研究旨在利用少量非标记校准数据建立一种可靠的警觉性估计方法。我们在设计的基于BCI的光标控制任务中进行了警觉性实验。我们在两个不同的日期分两次记录了18名参与者的脑电图（EEG）信号。然后，我们提出了一种用于警觉性估计的对比度细粒度域自适应网络（CFGDAN）。在这里，我们建立了一个自适应图卷积网络（GCN），将不同域的脑电图数据投射到一个共同的空间。设计了细粒度特征对齐机制，以在脑电图通道级别对不同域的特征分布进行加权和对齐，并开发了对比信息保存模块，以在特征对齐过程中保存有用的目标特定信息。实验结果表明，在我们的 BCI 警戒数据集和 SEED-VIG 数据集中，所提出的 CFGDAN 优于同类方法。此外，可视化结果也证明了所设计的特征配准机制的有效性。这些结果表明了我们的方法在警觉性估计方面的有效性。我们的研究有助于减少校准工作，提高警觉性估计方法的实际应用潜力。

{"title":"Contrastive fine-grained domain adaptation network for EEG-based vigilance estimation.","authors":"Kangning Wang, Wei Wei, Weibo Yi, Shuang Qiu, Huiguang He, Minpeng Xu, Dong Ming","doi":"10.1016/j.neunet.2024.106617","DOIUrl":"10.1016/j.neunet.2024.106617","url":null,"abstract":"<p><p>Vigilance state is crucial for the effective performance of users in brain-computer interface (BCI) systems. Most vigilance estimation methods rely on a large amount of labeled data to train a satisfactory model for the specific subject, which limits the practical application of the methods. This study aimed to build a reliable vigilance estimation method using a small amount of unlabeled calibration data. We conducted a vigilance experiment in the designed BCI-based cursor-control task. Electroencephalogram (EEG) signals of eighteen participants were recorded in two sessions on two different days. And, we proposed a contrastive fine-grained domain adaptation network (CFGDAN) for vigilance estimation. Here, an adaptive graph convolution network (GCN) was built to project the EEG data of different domains into a common space. The fine-grained feature alignment mechanism was designed to weight and align the feature distributions across domains at the EEG channel level, and the contrastive information preservation module was developed to preserve the useful target-specific information during the feature alignment. The experimental results show that the proposed CFGDAN outperforms the compared methods in our BCI vigilance dataset and SEED-VIG dataset. Moreover, the visualization results demonstrate the efficacy of the designed feature alignment mechanisms. These results indicate the effectiveness of our method for vigilance estimation. Our study is helpful for reducing calibration efforts and promoting the practical application potential of vigilance estimation methods.</p>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"179 ","pages":"106617"},"PeriodicalIF":6.0,"publicationDate":"2024-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142057086","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

GradToken: Decoupling tokens with class-aware gradient for visual explanation of Transformer network GradToken：用类感知梯度解耦标记，对变形网络进行可视化解释。

IF 6 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Neural Networks

Pub Date : 2024-11-01 DOI: 10.1016/j.neunet.2024.106837

Lin Cheng , Yanjie Liang , Yang Lu , Yiu-ming Cheung

Transformer networks have been widely used in the fields of computer vision, natural language processing, graph-structured data analysis, etc. Subsequently, explanations of Transformer play a key role in helping humans understand and analyze its decision-making and working mechanism, thereby improving the trustworthiness in its real-world applications. However, it is difficult to apply the existing explanation methods for convolutional neural networks to Transformer networks, due to the significant differences between their structures. How to design a specific and effective explanation method for Transformer poses a challenge in the explanation area. To address this challenge, we first analyze the semantic coupling problem of attention weight matrices in Transformer, which puts obstacles in providing distinctive explanations for different categories of targets. Then, we propose a gradient-decoupling-based token relevance method (i.e., GradToken) for the visual explanation of Transformer’s predictions. GradToken exploits the class-aware gradient to decouple the tangled semantics in the class token to the semantics corresponding to each category. GradToken further leverages the relations between the class token and spatial tokens to generate relevance maps. As a result, the visual explanation results generated by GradToken can effectively focus on the regions of selected targets. Extensive quantitative and qualitative experiments are conducted to verify the validity and reliability of the proposed method.

变换器网络已被广泛应用于计算机视觉、自然语言处理、图结构数据分析等领域。因此，对变形金刚的解释在帮助人类理解和分析其决策和工作机制方面起着关键作用，从而提高其在实际应用中的可信度。然而，由于卷积神经网络与变形金刚网络在结构上存在显著差异，现有的卷积神经网络解释方法很难应用于变形金刚网络。如何为 Transformer 设计一种具体而有效的解释方法是解释领域的一个难题。为了应对这一挑战，我们首先分析了 Transformer 中注意力权重矩阵的语义耦合问题，这一问题为针对不同类别的目标提供独特的解释设置了障碍。然后，我们提出了一种基于梯度解耦的标记相关性方法（即 GradToken），用于 Transformer 预测的视觉解释。GradToken 利用类别感知梯度，将类别标记中纠缠不清的语义与每个类别对应的语义解耦。GradToken 进一步利用类别标记和空间标记之间的关系生成相关性地图。因此，GradToken 生成的可视化解释结果可以有效地聚焦于选定目标的区域。为了验证所提方法的有效性和可靠性，我们进行了广泛的定量和定性实验。

{"title":"GradToken: Decoupling tokens with class-aware gradient for visual explanation of Transformer network","authors":"Lin Cheng , Yanjie Liang , Yang Lu , Yiu-ming Cheung","doi":"10.1016/j.neunet.2024.106837","DOIUrl":"10.1016/j.neunet.2024.106837","url":null,"abstract":"<div><div>Transformer networks have been widely used in the fields of computer vision, natural language processing, graph-structured data analysis, etc. Subsequently, explanations of Transformer play a key role in helping humans understand and analyze its decision-making and working mechanism, thereby improving the trustworthiness in its real-world applications. However, it is difficult to apply the existing explanation methods for convolutional neural networks to Transformer networks, due to the significant differences between their structures. How to design a specific and effective explanation method for Transformer poses a challenge in the explanation area. To address this challenge, we first analyze the semantic coupling problem of attention weight matrices in Transformer, which puts obstacles in providing distinctive explanations for different categories of targets. Then, we propose a gradient-decoupling-based token relevance method (i.e., GradToken) for the visual explanation of Transformer’s predictions. GradToken exploits the class-aware gradient to decouple the tangled semantics in the class token to the semantics corresponding to each category. GradToken further leverages the relations between the class token and spatial tokens to generate relevance maps. As a result, the visual explanation results generated by GradToken can effectively focus on the regions of selected targets. Extensive quantitative and qualitative experiments are conducted to verify the validity and reliability of the proposed method.</div></div>","PeriodicalId":49763,"journal":{"name":"Neural Networks","volume":"181 ","pages":"Article 106837"},"PeriodicalIF":6.0,"publicationDate":"2024-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142631325","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