Adversarial graph node classification based on unsupervised learning and optimized loss functions

3区计算机科学 Q1 Computer Science Journal of Ambient Intelligence and Humanized Computing Pub Date : 2024-03-11 DOI:10.1007/s12652-024-04768-0

{"title":"Adversarial graph node classification based on unsupervised learning and optimized loss functions","authors":"","doi":"10.1007/s12652-024-04768-0","DOIUrl":null,"url":null,"abstract":"<h3>Abstract</h3> <p>The research field of this paper is unsupervised learning in machine learning, aiming to address the problem of how to simultaneously resist feature attacks and improve model classification performance in unsupervised learning. For this purpose, this paper proposes a method to add an optimized loss function after the graph encoding and representation stage. When the samples are relatively balanced, we choose the cross-entropy loss function for classification. When difficult-to-classify samples appear, an optimized Focal Loss*() function is used to adjust the weights of these samples, to solve the problem of imbalanced positive and negative samples during training. The developed method achieved superior performance accuracy with the values of 0.721 on the Cora dataset, 0.598 on the Citeseer dataset,0.862 on the Polblogs dataset. Moreover, the testing accuracy value achieved by optimized model is 0.745, 0.627, 0.892 on the three benchmark datasets, respectively. Experimental results show that the proposed method effectively improves the robustness of adversarial training models in downstream tasks and reduces potential interference with original data. All the test results are validated with the k-fold cross validation method in order to make an assessment of the generalizability of these results.</p>","PeriodicalId":14959,"journal":{"name":"Journal of Ambient Intelligence and Humanized Computing","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Ambient Intelligence and Humanized Computing","FirstCategoryId":"94","ListUrlMain":"https://doi.org/10.1007/s12652-024-04768-0","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Computer Science","Score":null,"Total":0}

引用次数: 0

Abstract

The research field of this paper is unsupervised learning in machine learning, aiming to address the problem of how to simultaneously resist feature attacks and improve model classification performance in unsupervised learning. For this purpose, this paper proposes a method to add an optimized loss function after the graph encoding and representation stage. When the samples are relatively balanced, we choose the cross-entropy loss function for classification. When difficult-to-classify samples appear, an optimized Focal Loss*() function is used to adjust the weights of these samples, to solve the problem of imbalanced positive and negative samples during training. The developed method achieved superior performance accuracy with the values of 0.721 on the Cora dataset, 0.598 on the Citeseer dataset,0.862 on the Polblogs dataset. Moreover, the testing accuracy value achieved by optimized model is 0.745, 0.627, 0.892 on the three benchmark datasets, respectively. Experimental results show that the proposed method effectively improves the robustness of adversarial training models in downstream tasks and reduces potential interference with original data. All the test results are validated with the k-fold cross validation method in order to make an assessment of the generalizability of these results.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

基于无监督学习和优化损失函数的对抗性图节点分类

摘要本文的研究领域是机器学习中的无监督学习，旨在解决无监督学习中如何同时抵御特征攻击和提高模型分类性能的问题。为此，本文提出了一种在图编码和表示阶段之后添加优化损失函数的方法。当样本相对均衡时，我们选择交叉熵损失函数进行分类。当出现难以分类的样本时，则使用优化的 Focal Loss*() 函数来调整这些样本的权重，以解决训练过程中正负样本不平衡的问题。所开发的方法在 Cora 数据集、Citeseer 数据集和 Polblogs 数据集上的准确率分别达到了 0.721、0.598 和 0.862。此外，优化模型在三个基准数据集上的测试准确率分别为 0.745、0.627 和 0.892。实验结果表明，所提出的方法有效提高了对抗训练模型在下游任务中的鲁棒性，并减少了对原始数据的潜在干扰。所有测试结果都通过 k 倍交叉验证法进行了验证，以评估这些结果的普适性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

Journal of Ambient Intelligence and Humanized Computing COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCEC-COMPUTER SCIENCE, INFORMATION SYSTEMS

CiteScore

9.60

自引率

0.00%

发文量

854

期刊介绍： The purpose of JAIHC is to provide a high profile, leading edge forum for academics, industrial professionals, educators and policy makers involved in the field to contribute, to disseminate the most innovative researches and developments of all aspects of ambient intelligence and humanized computing, such as intelligent/smart objects, environments/spaces, and systems. The journal discusses various technical, safety, personal, social, physical, political, artistic and economic issues. The research topics covered by the journal are (but not limited to): Pervasive/Ubiquitous Computing and Applications Cognitive wireless sensor network Embedded Systems and Software Mobile Computing and Wireless Communications Next Generation Multimedia Systems Security, Privacy and Trust Service and Semantic Computing Advanced Networking Architectures Dependable, Reliable and Autonomic Computing Embedded Smart Agents Context awareness, social sensing and inference Multi modal interaction design Ergonomics and product prototyping Intelligent and self-organizing transportation networks & services Healthcare Systems Virtual Humans & Virtual Worlds Wearables sensors and actuators