Knowledge-Based Systems最新文献_第8页

Misalignment-resistant domain adaptive learning for one-stage object detection 用于单级物体检测的抗错位域自适应学习

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112605

Yunfei Bai , Chang Liu , Rui Yang , Xiaomao Li

Without consideration of task specificity, directly transforming domain adaptive pipelines from classification to one-stage detection tends to pose severer misalignments. These misalignments include: (1) Foreground misalignment that the domain discriminator obsessively concentrates on backgrounds since one-stage detectors do not contain proposals for instance-level discrimination. (2) Localization misalignment that domain-adaptive features supervised by the domain discriminator are not suitable for localization tasks, as the discriminator is a classifier in essence. To tackle these problems, we propose the Misalignment-Resistant Domain Adaption (MRDA) for one-stage detectors. Specifically, to alleviate foreground misalignment, a mask-based domain discriminator is proposed to perform instance-level discrimination by assigning the pixel-level domain labels based on instance-level masks. As for localization misalignment, a localization discriminator is introduced to learn domain-adaptive features for localization tasks. It employs an additional box-regression branch with an IoU loss to perform adversarial mutual supervision with the feature extractor. Comprehensive experiments demonstrate that our method effectively mitigates the misalignments and achieves state-of-the-art detection across multiple datasets.

如果不考虑任务的特殊性，直接将领域自适应管道从分类转换为单级检测，往往会造成更严重的错位。这些错位包括(1) 前景错位：由于单级检测器不包含实例级判别建议，因此领域判别器会过度关注背景。(2) 定位失准，领域判别器监督的领域自适应特征不适合定位任务，因为判别器本质上是一个分类器。为了解决这些问题，我们提出了针对单级检测器的抗错位域自适应（Misalignment-Resistant Domain Adaption，MRDA）。具体来说，为了减轻前景错位，我们提出了一种基于掩码的域判别器，通过基于实例级掩码分配像素级域标签来执行实例级判别。至于定位错位，则引入了一个定位判别器来学习定位任务的域自适应特征。它采用了额外的盒回归分支和 IoU 损失，与特征提取器一起执行对抗性相互监督。综合实验证明，我们的方法能有效缓解错位，并在多个数据集上实现了最先进的检测。

{"title":"Misalignment-resistant domain adaptive learning for one-stage object detection","authors":"Yunfei Bai , Chang Liu , Rui Yang , Xiaomao Li","doi":"10.1016/j.knosys.2024.112605","DOIUrl":"10.1016/j.knosys.2024.112605","url":null,"abstract":"<div><div>Without consideration of task specificity, directly transforming domain adaptive pipelines from classification to one-stage detection tends to pose severer misalignments. These misalignments include: (1) Foreground misalignment that the domain discriminator obsessively concentrates on backgrounds since one-stage detectors do not contain proposals for instance-level discrimination. (2) Localization misalignment that domain-adaptive features supervised by the domain discriminator are not suitable for localization tasks, as the discriminator is a classifier in essence. To tackle these problems, we propose the <strong>Misalignment-Resistant Domain Adaption</strong> (MRDA) for one-stage detectors. Specifically, to alleviate foreground misalignment, a mask-based domain discriminator is proposed to perform instance-level discrimination by assigning the pixel-level domain labels based on instance-level masks. As for localization misalignment, a localization discriminator is introduced to learn domain-adaptive features for localization tasks. It employs an additional box-regression branch with an IoU loss to perform adversarial mutual supervision with the feature extractor. Comprehensive experiments demonstrate that our method effectively mitigates the misalignments and achieves state-of-the-art detection across multiple datasets.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529487","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

Self-labeling in multivariate causality and quantification for adaptive machine learning 多元因果关系中的自标注和自适应机器学习的量化

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112595

Yutian Ren, Aaron Haohua Yen, G.P. Li

Adaptive machine learning (ML) aims to allow ML models to adapt to ever-changing environments with potential concept drift after model deployment. Traditionally, adaptive ML requires a new dataset to be manually labeled to tailor deployed models to altered data distributions. Recently, an interactive causality based self-labeling method was proposed to autonomously associate causally related data streams for domain adaptation, showing promising results compared to traditional feature similarity-based semi-supervised learning. Several unanswered research questions remain, including self-labeling’s compatibility with multivariate causality and the quantitative analysis of the auxiliary models used in the self-labeling. The auxiliary models, the interaction time model (ITM) and the effect state detector (ESD), are vital to the success of self-labeling. This paper further develops the self-labeling framework and its theoretical foundations to address these research questions. A framework for the application of self-labeling to multivariate causal graphs is proposed using four basic causal relationships, and the impact of non-ideal ITM and ESD performance is analyzed. A simulated experiment is conducted based on a multivariate causal graph, validating the proposed theory.

自适应机器学习（ML）旨在让机器学习模型适应不断变化的环境，并在模型部署后适应潜在的概念漂移。传统上，自适应 ML 需要对新数据集进行人工标注，以使部署的模型适应已改变的数据分布。最近，有人提出了一种基于因果关系的交互式自标注方法，可自主关联因果关系相关的数据流以进行领域适应性学习，与传统的基于特征相似性的半监督学习相比，这种方法显示出了良好的效果。目前仍有几个未解答的研究问题，包括自标注与多元因果关系的兼容性以及对自标注中使用的辅助模型的定量分析。辅助模型，即交互时间模型（ITM）和效应状态检测器（ESD），对自我标记的成功至关重要。本文进一步发展了自我标记框架及其理论基础，以解决这些研究问题。利用四种基本因果关系，提出了将自标记应用于多元因果图的框架，并分析了非理想 ITM 和 ESD 性能的影响。基于多变量因果图进行了模拟实验，验证了所提出的理论。

{"title":"Self-labeling in multivariate causality and quantification for adaptive machine learning","authors":"Yutian Ren, Aaron Haohua Yen, G.P. Li","doi":"10.1016/j.knosys.2024.112595","DOIUrl":"10.1016/j.knosys.2024.112595","url":null,"abstract":"<div><div>Adaptive machine learning (ML) aims to allow ML models to adapt to ever-changing environments with potential concept drift after model deployment. Traditionally, adaptive ML requires a new dataset to be manually labeled to tailor deployed models to altered data distributions. Recently, an interactive causality based self-labeling method was proposed to autonomously associate causally related data streams for domain adaptation, showing promising results compared to traditional feature similarity-based semi-supervised learning. Several unanswered research questions remain, including self-labeling’s compatibility with multivariate causality and the quantitative analysis of the auxiliary models used in the self-labeling. The auxiliary models, the interaction time model (ITM) and the effect state detector (ESD), are vital to the success of self-labeling. This paper further develops the self-labeling framework and its theoretical foundations to address these research questions. A framework for the application of self-labeling to multivariate causal graphs is proposed using four basic causal relationships, and the impact of non-ideal ITM and ESD performance is analyzed. A simulated experiment is conducted based on a multivariate causal graph, validating the proposed theory.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529372","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

Enhancing land cover classification via deep ensemble network 通过深度集合网络加强土地覆被分类

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112611

Muhammad Fayaz , L. Minh Dang , Hyeonjoon Moon

The rapid adoption of drones has transformed industries such as agriculture, environmental monitoring, surveillance, and disaster management by enabling more efficient data collection and analysis. However, existing UAV-based image scene classification techniques face limitations, particularly in handling dynamic scenes, varying environmental conditions, and accurately identifying small or partially obscured objects. These challenges necessitate more advanced and robust methods for land cover classification. In response, this study explores ensemble learning (EL) as a powerful alternative to traditional machine learning approaches. By integrating predictions from multiple models, EL enhances accuracy, precision, and robustness in UAV-based land use and land cover classification. This research introduces a two-phase approach combining data preprocessing with feature extraction using three advanced ensemble models DenseNet201, EfficientNetV2S, and Xception employing transfer learning. These models were selected based on their higher performance during preliminary evaluations. Furthermore, a soft attention mechanism is incorporated into the ensembled network to optimize feature selection, resulting in improved classification outcomes. The proposed model achieved an accuracy of 97 %, precision of 96 %, recall of 96 %, and an F1-score of 97 % on UAV image datasets. Comparative analysis reveals a 4.2 % accuracy improvement with the ensembled models and a 1 % boost with the advanced hybrid models. This work significantly advances UAV image scene classification, offering a practical solution to enhance decision-making precision in various applications. The ensemble system demonstrates its effectiveness in remote sensing applications, especially in land cover analysis across diverse geographical and environmental settings.

无人机的快速应用改变了农业、环境监测、监控和灾害管理等行业，实现了更高效的数据收集和分析。然而，现有的基于无人机的图像场景分类技术面临着局限性，尤其是在处理动态场景、不同环境条件以及准确识别小型或部分遮挡物体方面。面对这些挑战，有必要采用更先进、更稳健的方法来进行土地覆被分类。为此，本研究探索了集合学习（EL），将其作为传统机器学习方法的有力替代方案。通过整合多个模型的预测结果，集合学习提高了基于无人机的土地利用和土地覆被分类的准确性、精确性和稳健性。本研究介绍了一种两阶段方法，将数据预处理与特征提取相结合，并使用三种先进的集合模型 DenseNet201、EfficientNetV2S 和采用迁移学习的 Xception。之所以选择这些模型，是因为它们在初步评估中表现较好。此外，还在集合网络中加入了软关注机制，以优化特征选择，从而改善分类结果。所提出的模型在无人机图像数据集上的准确率达到 97%，精确率达到 96%，召回率达到 96%，F1 分数达到 97%。对比分析表明，集合模型的准确率提高了 4.2%，而高级混合模型的准确率提高了 1%。这项工作极大地推动了无人机图像场景分类，为提高各种应用中的决策精度提供了实用的解决方案。集合系统证明了其在遥感应用中的有效性，特别是在不同地理和环境背景下的土地覆盖分析中。

{"title":"Enhancing land cover classification via deep ensemble network","authors":"Muhammad Fayaz , L. Minh Dang , Hyeonjoon Moon","doi":"10.1016/j.knosys.2024.112611","DOIUrl":"10.1016/j.knosys.2024.112611","url":null,"abstract":"<div><div>The rapid adoption of drones has transformed industries such as agriculture, environmental monitoring, surveillance, and disaster management by enabling more efficient data collection and analysis. However, existing UAV-based image scene classification techniques face limitations, particularly in handling dynamic scenes, varying environmental conditions, and accurately identifying small or partially obscured objects. These challenges necessitate more advanced and robust methods for land cover classification. In response, this study explores ensemble learning (EL) as a powerful alternative to traditional machine learning approaches. By integrating predictions from multiple models, EL enhances accuracy, precision, and robustness in UAV-based land use and land cover classification. This research introduces a two-phase approach combining data preprocessing with feature extraction using three advanced ensemble models DenseNet201, EfficientNetV2S, and Xception employing transfer learning. These models were selected based on their higher performance during preliminary evaluations. Furthermore, a soft attention mechanism is incorporated into the ensembled network to optimize feature selection, resulting in improved classification outcomes. The proposed model achieved an accuracy of 97 %, precision of 96 %, recall of 96 %, and an F1-score of 97 % on UAV image datasets. Comparative analysis reveals a 4.2 % accuracy improvement with the ensembled models and a 1 % boost with the advanced hybrid models. This work significantly advances UAV image scene classification, offering a practical solution to enhance decision-making precision in various applications. The ensemble system demonstrates its effectiveness in remote sensing applications, especially in land cover analysis across diverse geographical and environmental settings.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529375","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

Modeling group-level public sentiment in social networks through topic and role enhancement 通过增强话题和角色来模拟社交网络中的群体级公众情绪

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112594

Ruwen Zhang , Bo Liu , Jiuxin Cao , Hantao Zhao , Xuheng Sun , Yan Liu , Xiangguo Sun

Public sentiment within social networks exerts a profound influence on societal dynamics, underscoring the increasing demand for accurate public opinion prediction. Most existing methods predominantly measure sentiment by quantifying user sentiments individually, overlooking group-level factors that crucially contribute to public sentiment. Thus, based on our finding that public sentiment is primarily shaped by user-group interactions and their interplay with evolving topics, we innovatively model the forming process of public sentiment at the group level. In this paper, we propose the Topic and Role Enhanced Group-level Public Sentiment Prediction model (TRESP), capturing the intricate interplay among sentiment, topic, and role. Specifically, an LSTM neural network is firstly leveraged to trace the temporal correlations between topics and sentiment shifts, yielding a topic-informed content sentiment representation. Subsequently, a specially crafted hierarchical attention network captures social and role attributes, representing the overarching social group environment. Finally, we predict future public sentiment by merging the derived group sentiment representation with the group social representation, demonstrating a holistic insight into the sentiment trajectory. Extensive experiments were conducted on two real-world datasets of over 30,000 tweets collected from more than 14,000 users to validate our model. Notably, our model significantly outperforms the state-of-the-art approaches in public sentiment prediction, indicating the importance and effectiveness of encapsulating interactions both within and among user subgroups.

社交网络中的公众情绪对社会动态有着深远的影响，因此对准确舆论预测的需求与日俱增。现有的大多数方法主要通过量化用户的个人情绪来衡量情绪，而忽略了对公众情绪起关键作用的群体层面的因素。因此，基于我们发现公众情绪主要是由用户-群体互动及其与不断演变的话题之间的相互作用形成的这一发现，我们创新性地从群体层面对公众情绪的形成过程进行了建模。在本文中，我们提出了话题和角色增强型群体级公众情绪预测模型（TRESP），以捕捉情绪、话题和角色之间错综复杂的相互作用。具体来说，我们首先利用 LSTM 神经网络来追踪话题与情感转变之间的时间相关性，从而得到一个以话题为基础的内容情感表征。随后，一个专门设计的分层注意力网络会捕捉社会和角色属性，代表社会群体的总体环境。最后，我们将得出的群体情感表征与群体社会表征合并，预测未来的公众情感，从而展现出对情感轨迹的整体洞察力。为了验证我们的模型，我们在两个真实世界的数据集上进行了广泛的实验，这些数据集包含了从超过 14,000 名用户那里收集的 30,000 多条推文。值得注意的是，在公众情绪预测方面，我们的模型明显优于最先进的方法，这表明了封装用户子群内部和之间互动的重要性和有效性。

{"title":"Modeling group-level public sentiment in social networks through topic and role enhancement","authors":"Ruwen Zhang , Bo Liu , Jiuxin Cao , Hantao Zhao , Xuheng Sun , Yan Liu , Xiangguo Sun","doi":"10.1016/j.knosys.2024.112594","DOIUrl":"10.1016/j.knosys.2024.112594","url":null,"abstract":"<div><div>Public sentiment within social networks exerts a profound influence on societal dynamics, underscoring the increasing demand for accurate public opinion prediction. Most existing methods predominantly measure sentiment by quantifying user sentiments individually, overlooking group-level factors that crucially contribute to public sentiment. Thus, based on our finding that public sentiment is primarily shaped by user-group interactions and their interplay with evolving topics, we innovatively model the forming process of public sentiment at the group level. In this paper, we propose the Topic and Role Enhanced Group-level Public Sentiment Prediction model (TRESP), capturing the intricate interplay among sentiment, topic, and role. Specifically, an LSTM neural network is firstly leveraged to trace the temporal correlations between topics and sentiment shifts, yielding a topic-informed content sentiment representation. Subsequently, a specially crafted hierarchical attention network captures social and role attributes, representing the overarching social group environment. Finally, we predict future public sentiment by merging the derived group sentiment representation with the group social representation, demonstrating a holistic insight into the sentiment trajectory. Extensive experiments were conducted on two real-world datasets of over 30,000 tweets collected from more than 14,000 users to validate our model. Notably, our model significantly outperforms the state-of-the-art approaches in public sentiment prediction, indicating the importance and effectiveness of encapsulating interactions both within and among user subgroups.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142446719","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

Enhancing the security in IoT and IIoT networks: An intrusion detection scheme leveraging deep transfer learning 增强物联网和 IIoT 网络的安全性：利用深度迁移学习的入侵检测方案

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112614

Basharat Ahmad , Zhaoliang Wu , Yongfeng Huang , Sadaqat Ur Rehman

The Internet of Things (IoT) networks, which are defined by their interconnected devices and data streams are an expanding attack surface for cyber adversaries. Industrial Internet of Things (IIoT) is a subset of IoT and has significant importance in-terms of security. Robust intrusion detection systems (IDS) are essential for protecting these critical infrastructures. Our research suggests a novel approach to the detection of anomalies in IoT and IIoT networks that leverages the capabilities of deep transfer learning. Our methodology begins with the EdgeIIoT dataset, which serves as the basis for our data analysis. We convert the data into an appropriate image format to enable Convolutional Neural Network (CNN)-based processing. The hyper-parameters of individual machine learning models are subsequently optimized using a Random Search algorithm. This optimization phase optimizes the performance of each model by modifying the hyper-parameters that are unique to the learning algorithms. The performance of each model is meticulously assessed subsequent to hyper-parameter optimization. The top-performing models are subsequently, strategically selected and combined using the ensemble technique. The IDS scheme’s overall detection accuracy and generalizability are improved by the integration of strengths from multiple models. The proposed scheme demonstrates significant effectiveness in identifying a broad spectrum of attacks, encompassing a total of 14 distinct attack types. This comprehensive detection capability contributes to a more secure and resilient IoT ecosystem. Furthermore, application of quantization to our best models reduces resource utilization significantly without compromising accuracy.

物联网（IoT）网络由相互连接的设备和数据流构成，是网络对手不断扩大的攻击面。工业物联网（IIoT）是物联网的一个子集，在安全方面具有重要意义。强大的入侵检测系统（IDS）对于保护这些关键基础设施至关重要。我们的研究提出了一种利用深度迁移学习能力检测物联网和 IIoT 网络异常的新方法。我们的方法从 EdgeIIoT 数据集开始，该数据集是我们进行数据分析的基础。我们将数据转换成适当的图像格式，以便进行基于卷积神经网络（CNN）的处理。随后使用随机搜索算法对各个机器学习模型的超参数进行优化。该优化阶段通过修改学习算法特有的超参数来优化每个模型的性能。超参数优化后，每个模型的性能都会得到细致的评估。随后，利用集合技术战略性地选择并组合性能最佳的模型。通过整合多个模型的优势，IDS 方案的整体检测精度和通用性都得到了提高。所提出的方案在识别各种攻击（共包括 14 种不同的攻击类型）方面效果显著。这种全面的检测能力有助于建立一个更安全、更有弹性的物联网生态系统。此外，在我们的最佳模型中应用量化技术大大降低了资源利用率，同时又不影响准确性。

{"title":"Enhancing the security in IoT and IIoT networks: An intrusion detection scheme leveraging deep transfer learning","authors":"Basharat Ahmad , Zhaoliang Wu , Yongfeng Huang , Sadaqat Ur Rehman","doi":"10.1016/j.knosys.2024.112614","DOIUrl":"10.1016/j.knosys.2024.112614","url":null,"abstract":"<div><div>The Internet of Things (IoT) networks, which are defined by their interconnected devices and data streams are an expanding attack surface for cyber adversaries. Industrial Internet of Things (IIoT) is a subset of IoT and has significant importance in-terms of security. Robust intrusion detection systems (IDS) are essential for protecting these critical infrastructures. Our research suggests a novel approach to the detection of anomalies in IoT and IIoT networks that leverages the capabilities of deep transfer learning. Our methodology begins with the EdgeIIoT dataset, which serves as the basis for our data analysis. We convert the data into an appropriate image format to enable Convolutional Neural Network (CNN)-based processing. The hyper-parameters of individual machine learning models are subsequently optimized using a Random Search algorithm. This optimization phase optimizes the performance of each model by modifying the hyper-parameters that are unique to the learning algorithms. The performance of each model is meticulously assessed subsequent to hyper-parameter optimization. The top-performing models are subsequently, strategically selected and combined using the ensemble technique. The IDS scheme’s overall detection accuracy and generalizability are improved by the integration of strengths from multiple models. The proposed scheme demonstrates significant effectiveness in identifying a broad spectrum of attacks, encompassing a total of 14 distinct attack types. This comprehensive detection capability contributes to a more secure and resilient IoT ecosystem. Furthermore, application of quantization to our best models reduces resource utilization significantly without compromising accuracy.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529370","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

MAS-DGAT-Net: A dynamic graph attention network with multibranch feature extraction and staged fusion for EEG emotion recognition MAS-DGAT-Net：用于脑电图情感识别的多分支特征提取和分阶段融合的动态图注意网络

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112599

Shuaiqi Liu , Xinrui Wang , Mingqi Jiang , Yanling An , Zhihui Gu , Bing Li , Yudong Zhang

In recent years, with the rise of deep learning technologies, EEG-based emotion recognition has garnered significant attention. However, most existing methods tend to focus on the spatiotemporal information of EEG signals while overlooking the potential topological information of brain regions. To address this issue, this paper proposes a dynamic graph attention network with multi-branch feature extraction and staged fusion (MAS-DGAT-Net), which integrates graph convolutional neural networks (GCN) for EEG emotion recognition. Specifically, the differential entropy (DE) features of EEG signals are first reconstructed into a correlation matrix using the Spearman correlation coefficient. Then, the brain-region connectivity-feature extraction (BCFE) module is employed to capture the brain connectivity features associated with emotional activation states. Meanwhile, this paper introduces a dual-branch cross-fusion feature extraction (CFFE) module, which consists of an attention-based cross-fusion feature extraction branch (A-CFFEB) and a cross-fusion feature extraction branch (CFFEB). A-CFFEB efficiently extracts key channel-frequency information from EEG features by using an attention mechanism and then fuses it with the output features from the BCFE. The fused features are subsequently input into the proposed dynamic graph attention module with a broad learning system (DGAT-BLS) to mine the brain connectivity feature information further. Finally, the deep features output by DGAT-BLS and CFFEB are combined for emotion classification. The proposed algorithm has been experimentally validated on SEED, SEED-IV, and DEAP datasets in subject-dependent and subject-independent settings, with the results confirming the model's effectiveness. The source code is publicly available at: https://github.com/cvmdsp/MAS-DGAT-Net

近年来，随着深度学习技术的兴起，基于脑电图的情感识别受到了广泛关注。然而，现有的大多数方法往往只关注脑电信号的时空信息，而忽略了脑区潜在的拓扑信息。为解决这一问题，本文提出了一种具有多分支特征提取和分阶段融合功能的动态图注意力网络（MAS-DGAT-Net），它将图卷积神经网络（GCN）整合到脑电图情绪识别中。具体来说，首先使用斯皮尔曼相关系数将脑电信号的差分熵（DE）特征重建为相关矩阵。然后，利用脑区连接特征提取（BCFE）模块捕捉与情绪激活状态相关的脑连接特征。同时，本文引入了双分支交叉融合特征提取（CFFE）模块，该模块由基于注意力的交叉融合特征提取分支（A-CFFEB）和交叉融合特征提取分支（CFFEB）组成。A-CFFEB 利用注意力机制从脑电图特征中有效提取关键信道频率信息，然后将其与 BCFE 的输出特性融合。融合后的特征随后被输入带有广泛学习系统的动态图注意模块（DGAT-BLS），以进一步挖掘大脑连接特征信息。最后，将 DGAT-BLS 和 CFFEB 输出的深度特征结合起来进行情绪分类。所提出的算法已在 SEED、SEED-IV 和 DEAP 数据集上进行了实验验证，包括依赖主体和不依赖主体两种设置，结果证实了模型的有效性。源代码可在以下网址公开获取： https://github.com/cvmdsp/MAS-DGAT-Net

{"title":"MAS-DGAT-Net: A dynamic graph attention network with multibranch feature extraction and staged fusion for EEG emotion recognition","authors":"Shuaiqi Liu , Xinrui Wang , Mingqi Jiang , Yanling An , Zhihui Gu , Bing Li , Yudong Zhang","doi":"10.1016/j.knosys.2024.112599","DOIUrl":"10.1016/j.knosys.2024.112599","url":null,"abstract":"<div><div>In recent years, with the rise of deep learning technologies, EEG-based emotion recognition has garnered significant attention. However, most existing methods tend to focus on the spatiotemporal information of EEG signals while overlooking the potential topological information of brain regions. To address this issue, this paper proposes a dynamic graph attention network with multi-branch feature extraction and staged fusion (MAS-DGAT-Net), which integrates graph convolutional neural networks (GCN) for EEG emotion recognition. Specifically, the differential entropy (DE) features of EEG signals are first reconstructed into a correlation matrix using the Spearman correlation coefficient. Then, the brain-region connectivity-feature extraction (BCFE) module is employed to capture the brain connectivity features associated with emotional activation states. Meanwhile, this paper introduces a dual-branch cross-fusion feature extraction (CFFE) module, which consists of an attention-based cross-fusion feature extraction branch (A-CFFEB) and a cross-fusion feature extraction branch (CFFEB). A-CFFEB efficiently extracts key channel-frequency information from EEG features by using an attention mechanism and then fuses it with the output features from the BCFE. The fused features are subsequently input into the proposed dynamic graph attention module with a broad learning system (DGAT-BLS) to mine the brain connectivity feature information further. Finally, the deep features output by DGAT-BLS and CFFEB are combined for emotion classification. The proposed algorithm has been experimentally validated on SEED, SEED-IV, and DEAP datasets in subject-dependent and subject-independent settings, with the results confirming the model's effectiveness. The source code is publicly available at: <span><span>https://github.com/cvmdsp/MAS-DGAT-Net</span><svg><path></path></svg></span></div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529410","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

GAN-based statistical process control for the time series data 基于 GAN 的时间序列数据统计过程控制

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112613

Yu-Jeong Cheon , Wook-Yeon Hwang

The cumulative sum (CUSUM) control chart and the multivariate anomaly detection with the generative adversarial network (MAD-GAN) were compared for monitoring the time series data. However, the control boundaries constructed in terms of the one-class classification with only the normal data for the training phase are inappropriate for the test phase because the normal data and the abnormal data should be classified for the test phase. In this regard, we first propose this GAN-based statistical process control (SPC) framework to compare them in terms of detecting the process mean shift based on the perspective of SPC. Second, we propose the residual MAD-GAN in order to improve the detection performance. Third, we develop the loss function of the MAD-GAN. Finally, we find that the maximum mean discrepancy (MMD) as well as the nash equilibrium is useful for the MAD-GAN. Our experiments demonstrate that the residual MAD-GAN is more effective than the residual CUSUM control chart in terms of the run lengths for the time series data. Therefore, we propose SPC practitioners to consider the residual MAD-GAN for detecting the process mean shift in time series data.

在监测时间序列数据时，比较了累积和（CUSUM）控制图和生成式对抗网络（MAD-GAN）多变量异常检测。然而，在训练阶段仅使用正常数据进行单类分类而构建的控制边界并不适用于测试阶段，因为在测试阶段需要对正常数据和异常数据进行分类。为此，我们首先提出了基于 GAN 的统计过程控制 (SPC) 框架，从 SPC 的角度对它们在检测过程均值偏移方面进行比较。其次，我们提出了残差 MAD-GAN 以提高检测性能。第三，我们开发了 MAD-GAN 的损失函数。最后，我们发现最大均值差异（MMD）和纳什均衡对 MAD-GAN 非常有用。我们的实验证明，就时间序列数据的运行长度而言，残差 MAD-GAN 比残差 CUSUM 控制图更有效。因此，我们建议 SPC 从业人员考虑使用残差 MAD-GAN 来检测时间序列数据中的过程均值偏移。

{"title":"GAN-based statistical process control for the time series data","authors":"Yu-Jeong Cheon , Wook-Yeon Hwang","doi":"10.1016/j.knosys.2024.112613","DOIUrl":"10.1016/j.knosys.2024.112613","url":null,"abstract":"<div><div>The cumulative sum (CUSUM) control chart and the multivariate anomaly detection with the generative adversarial network (MAD-GAN) were compared for monitoring the time series data. However, the control boundaries constructed in terms of the one-class classification with only the normal data for the training phase are inappropriate for the test phase because the normal data and the abnormal data should be classified for the test phase. In this regard, we first propose this GAN-based statistical process control (SPC) framework to compare them in terms of detecting the process mean shift based on the perspective of SPC. Second, we propose the residual MAD-GAN in order to improve the detection performance. Third, we develop the loss function of the MAD-GAN. Finally, we find that the maximum mean discrepancy (MMD) as well as the nash equilibrium is useful for the MAD-GAN. Our experiments demonstrate that the residual MAD-GAN is more effective than the residual CUSUM control chart in terms of the run lengths for the time series data. Therefore, we propose SPC practitioners to consider the residual MAD-GAN for detecting the process mean shift in time series data.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529485","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

Augmentation-aware self-supervised learning with conditioned projector 带条件投影仪的增强感知自我监督学习

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112572

Marcin Przewięźlikowski , Mateusz Pyla , Bartosz Zieliński , Bartłomiej Twardowski , Jacek Tabor , Marek Śmieja

Self-supervised learning (SSL) is a powerful technique for learning from unlabeled data. By learning to remain invariant to applied data augmentations, methods such as SimCLR and MoCo can reach quality on par with supervised approaches. However, this invariance may be detrimental for solving downstream tasks that depend on traits affected by augmentations used during pretraining, such as color. In this paper, we propose to foster sensitivity to such characteristics in the representation space by modifying the projector network, a common component of self-supervised architectures. Specifically, we supplement the projector with information about augmentations applied to images. For the projector to take advantage of this auxiliary conditioning when solving the SSL task, the feature extractor learns to preserve the augmentation information in its representations. Our approach, coined Conditional Augmentation-aware Self-supervised Learning (CASSLE), is directly applicable to typical joint-embedding SSL methods regardless of their objective functions. Moreover, it does not require major changes in the network architecture or prior knowledge of downstream tasks. In addition to an analysis of sensitivity towards different data augmentations, we conduct a series of experiments, which show that CASSLE improves over various SSL methods, reaching state-of-the-art performance in multiple downstream tasks. ¹ ² ³

自我监督学习（SSL）是一种从无标记数据中学习的强大技术。通过学习保持对应用数据增强的不变性，SimCLR 和 MoCo 等方法可以达到与监督方法相当的质量。然而，这种不变性可能不利于解决下游任务，因为这些任务依赖于受预训练时使用的增强所影响的特征，如颜色。在本文中，我们建议通过修改投影仪网络（自我监督架构的常见组件）来提高对表示空间中此类特征的敏感性。具体来说，我们为投影器补充了有关图像增强的信息。为了让投影仪在解决 SSL 任务时利用这种辅助条件，特征提取器要学会在其表征中保留增强信息。我们的方法被称为条件增强感知自监督学习（CASSLE），可直接应用于典型的联合嵌入式 SSL 方法，而无需考虑其目标函数。此外，它不需要对网络架构进行重大改动，也不需要事先了解下游任务。除了分析对不同数据增强的敏感性外，我们还进行了一系列实验，结果表明 CASSLE 比各种 SSL 方法都有改进，在多个下游任务中达到了最先进的性能。1 2 3

{"title":"Augmentation-aware self-supervised learning with conditioned projector","authors":"Marcin Przewięźlikowski , Mateusz Pyla , Bartosz Zieliński , Bartłomiej Twardowski , Jacek Tabor , Marek Śmieja","doi":"10.1016/j.knosys.2024.112572","DOIUrl":"10.1016/j.knosys.2024.112572","url":null,"abstract":"<div><div>Self-supervised learning (SSL) is a powerful technique for learning from unlabeled data. By learning to remain invariant to applied data augmentations, methods such as SimCLR and MoCo can reach quality on par with supervised approaches. However, this invariance may be detrimental for solving downstream tasks that depend on traits affected by augmentations used during pretraining, such as color. In this paper, we propose to foster sensitivity to such characteristics in the representation space by modifying the projector network, a common component of self-supervised architectures. Specifically, we supplement the projector with information about augmentations applied to images. For the projector to take advantage of this auxiliary conditioning when solving the SSL task, the feature extractor learns to preserve the augmentation information in its representations. Our approach, coined <strong>C</strong>onditional <strong>A</strong>ugmentation-aware <strong>S</strong>elf-<strong>s</strong>upervised <strong>Le</strong>arning (CASSLE), is directly applicable to typical joint-embedding SSL methods regardless of their objective functions. Moreover, it does not require major changes in the network architecture or prior knowledge of downstream tasks. In addition to an analysis of sensitivity towards different data augmentations, we conduct a series of experiments, which show that CASSLE improves over various SSL methods, reaching state-of-the-art performance in multiple downstream tasks. <span><span><sup>1</sup></span></span> <span><span><sup>2</sup></span></span> <span><span><sup>3</sup></span></span></div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529408","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

Automated multiple sclerosis progression rate computation of a patient from 2D FLAIR images with Rayleigh-Weibull-Fuzzy imaging and augmented morphing method 利用 Rayleigh-Weibull-Fuzzy 成像和增强变形法，从二维 FLAIR 图像自动计算一名患者的多发性硬化症进展率

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112580

Orcan Alpar , Ondrej Soukup , Pavel Ryska , Petr Paluska , Martin Valis , Ondrej Krejcar

Multiple sclerosis (MS) is a neurological demyelinating disorder affecting brain and spinal cord by attacking the myelin sheaths of nerves. Estimation of the volumetric changes in MS lesions is a challenging and specialized task which is executed and judged by medical experts. The change in the volume of the lesions provides crucial information on MS progression or regression by comparing the magnetic resonance images (MRI) taken in successive scans. However, visual comparison of the images, even with an expert eye, would not always lead to a conclusive decision nor a consensus on progression or regression. Therefore, we present an automated expert system for estimating MS progression rate by automatic lesion segmentation and volume estimation using two-dimensional MRIs, which is also adaptable to various parameters, slice thickness and increment. A clinical dataset is specially formed for this research which contains three sets of 135 MR images of an MS patient generated within approximately 23- and 6-month periods consecutively with identical device parameters. The lesions are segmented by a novel Rayleigh-Weibull-Fuzzy (RWF) imaging method based on the Nakagami distribution and specialized fuzzy 2-means. Subsequently, the segmentation module is trained to fit the ground truths images created by experts to achieve the highest dice score possible for a total number of 56 images containing lesions, which is found as 93.76 %. Afterwards, several imaginary image sequences are generated by augmented linear and nonlinear morphing for re-segmentation of imaginary lesions by RWF. Finally, we estimated the volumetric change between the first two MRI sequences to adjust the morphing module and to predict the progression rate of the lesions in time. The framework automatically selected the highest accuracy, which is 99.9 % in the training session and estimated the progression rate in the testing phase with 99.69 % accuracy, which are not achievable without augmented morphing methodology. For the first time in the literature, an automated framework could estimate the MS progression rate from the raw MR images, which is also the main innovation of this paper and the outputs would be beneficial for the experts working on this field.

多发性硬化症（MS）是一种神经系统脱髓鞘疾病，通过破坏神经的髓鞘影响大脑和脊髓。多发性硬化病灶体积变化的估算是一项具有挑战性的专业任务，需要由医学专家来执行和判断。通过比较连续扫描的磁共振图像（MRI），病灶体积的变化为多发性硬化症的进展或消退提供了重要信息。然而，即使是通过专家的肉眼对图像进行比较，也并不总能得出结论，或就进展或消退达成共识。因此，我们提出了一种自动专家系统，通过使用二维核磁共振成像进行自动病灶分割和体积估算，来估算多发性硬化症的进展率，该系统还能适应各种参数、切片厚度和增量。这项研究专门建立了一个临床数据集，其中包含一名多发性硬化症患者在大约 23 个月和 6 个月期间连续生成的三组 135 幅 MR 图像，设备参数完全相同。病变是通过一种基于中神分布和专门模糊 2 均值的新型 Rayleigh-Weibull-Fuzzy (RWF) 成像方法分割的。随后，对分割模块进行训练，以拟合专家创建的地面实况图像，从而在总共 56 幅包含病变的图像中获得尽可能高的骰子分数，结果发现骰子分数为 93.76%。然后，通过增强线性和非线性变形生成多个假想图像序列，以便利用 RWF 对假想病灶进行重新分割。最后，我们估算了前两个核磁共振成像序列之间的体积变化，以调整变形模块，并预测病变在时间上的进展率。该框架在训练阶段自动选择了准确率最高的 99.9%，在测试阶段估算病变进展率的准确率为 99.69%，而这些准确率在没有增强变形方法的情况下是无法实现的。这也是本文的主要创新之处，其结果将对该领域的专家有所帮助。

{"title":"Automated multiple sclerosis progression rate computation of a patient from 2D FLAIR images with Rayleigh-Weibull-Fuzzy imaging and augmented morphing method","authors":"Orcan Alpar , Ondrej Soukup , Pavel Ryska , Petr Paluska , Martin Valis , Ondrej Krejcar","doi":"10.1016/j.knosys.2024.112580","DOIUrl":"10.1016/j.knosys.2024.112580","url":null,"abstract":"<div><div>Multiple sclerosis (MS) is a neurological demyelinating disorder affecting brain and spinal cord by attacking the myelin sheaths of nerves. Estimation of the volumetric changes in MS lesions is a challenging and specialized task which is executed and judged by medical experts. The change in the volume of the lesions provides crucial information on MS progression or regression by comparing the magnetic resonance images (MRI) taken in successive scans. However, visual comparison of the images, even with an expert eye, would not always lead to a conclusive decision nor a consensus on progression or regression. Therefore, we present an automated expert system for estimating MS progression rate by automatic lesion segmentation and volume estimation using two-dimensional MRIs, which is also adaptable to various parameters, slice thickness and increment. A clinical dataset is specially formed for this research which contains three sets of 135 MR images of an MS patient generated within approximately 23- and 6-month periods consecutively with identical device parameters. The lesions are segmented by a novel Rayleigh-Weibull-Fuzzy (RWF) imaging method based on the Nakagami distribution and specialized fuzzy 2-means. Subsequently, the segmentation module is trained to fit the ground truths images created by experts to achieve the highest dice score possible for a total number of 56 images containing lesions, which is found as 93.76 %. Afterwards, several imaginary image sequences are generated by augmented linear and nonlinear morphing for re-segmentation of imaginary lesions by RWF. Finally, we estimated the volumetric change between the first two MRI sequences to adjust the morphing module and to predict the progression rate of the lesions in time. The framework automatically selected the highest accuracy, which is 99.9 % in the training session and estimated the progression rate in the testing phase with 99.69 % accuracy, which are not achievable without augmented morphing methodology. For the first time in the literature, an automated framework could estimate the MS progression rate from the raw MR images, which is also the main innovation of this paper and the outputs would be beneficial for the experts working on this field.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529409","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

An autoencoder-like deep NMF representation learning algorithm for clustering 用于聚类的类似自动编码器的深度 NMF 表示学习算法

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-10 DOI: 10.1016/j.knosys.2024.112597

Dexian Wang , Pengfei Zhang , Ping Deng , Qiaofeng Wu , Wei Chen , Tao Jiang , Wei Huang , Tianrui Li

Clustering plays a crucial role in the field of data mining, where deep non-negative matrix factorization (NMF) has attracted significant attention due to its effective data representation. However, deep matrix factorization based on autoencoder is typically constructed using multi-layer matrix factorization, which ignores nonlinear mapping and lacks learning rate to guide the update. To address these issues, this paper proposes an autoencoder-like deep NMF representation learning (ADNRL) algorithm for clustering. First, according to the principle of autoencoder, construct the objective function based on NMF. Then, decouple the elements in the matrix and apply the nonlinear activation function to enforce non-negative constraints on the elements. Subsequently, the gradient values corresponding to the elements update guided by the learning rate are transformed into the weight values. This weight values are combined with the activation function to construct the ADNRL deep network, and the objective function is minimized through the learning of the network. Finally, extensive experiments are conducted on eight datasets, and the results demonstrate the superior performance of ADNRL.

聚类在数据挖掘领域起着至关重要的作用，其中深度非负矩阵因式分解（NMF）因其有效的数据表示而备受关注。然而，基于自动编码器的深度矩阵因式分解通常使用多层矩阵因式分解来构建，忽略了非线性映射，并且缺乏学习率来指导更新。针对这些问题，本文提出了一种类自编码器的深度 NMF 表示学习（ADNRL）聚类算法。首先，根据自动编码器的原理，构建基于 NMF 的目标函数。然后，解耦矩阵中的元素，并应用非线性激活函数对元素执行非负约束。随后，由学习率引导的元素更新所对应的梯度值被转化为权值。该权值与激活函数相结合，构建出 ADNRL 深度网络，并通过网络的学习使目标函数最小化。最后，在八个数据集上进行了大量实验，结果证明了 ADNRL 的卓越性能。

{"title":"An autoencoder-like deep NMF representation learning algorithm for clustering","authors":"Dexian Wang , Pengfei Zhang , Ping Deng , Qiaofeng Wu , Wei Chen , Tao Jiang , Wei Huang , Tianrui Li","doi":"10.1016/j.knosys.2024.112597","DOIUrl":"10.1016/j.knosys.2024.112597","url":null,"abstract":"<div><div>Clustering plays a crucial role in the field of data mining, where deep non-negative matrix factorization (NMF) has attracted significant attention due to its effective data representation. However, deep matrix factorization based on autoencoder is typically constructed using multi-layer matrix factorization, which ignores nonlinear mapping and lacks learning rate to guide the update. To address these issues, this paper proposes an autoencoder-like deep NMF representation learning (ADNRL) algorithm for clustering. First, according to the principle of autoencoder, construct the objective function based on NMF. Then, decouple the elements in the matrix and apply the nonlinear activation function to enforce non-negative constraints on the elements. Subsequently, the gradient values corresponding to the elements update guided by the learning rate are transformed into the weight values. This weight values are combined with the activation function to construct the ADNRL deep network, and the objective function is minimized through the learning of the network. Finally, extensive experiments are conducted on eight datasets, and the results demonstrate the superior performance of ADNRL.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":null,"pages":null},"PeriodicalIF":7.2,"publicationDate":"2024-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142442341","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