[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121637-en":3,"doc-seo-121637-105":30,"detail-sidebar-cat-0-en-105":95},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121637,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1784791008015729253",8,"Research & Report","UnbiasedNets - A Dataset Diversification Framework for Robustness Bias Alleviation in Neural Networks","Trained neural network models can achieve high testing accuracy while still exhibiting a robustness bias: performance under noisy inputs can be disproportionately strong for certain output classes and weak for others. This paper attributes robustness bias to imbalanced training datasets where classes are not equally represented. The proposed UnbiasedNets framework uses K-means clustering together with the network’s noise tolerance to diversify training data and generate more balanced datasets. Experiments on real-world datasets for both binary and multi-label classifiers show that UnbiasedNets reduces robustness bias and improves results versus existing balancing tools.","arXiv :2302 . 12538v2 [ cs .LG] 13 Mar 2023  \nSpringer Machine Learning Journal  \nThis version of the article has been accepted for publication, after peer review. The 􀀌nal version is available online at:  \n[http://dx.doi.org/10.1007/s10994-023-06314-z](http://dx.doi.org/10.1007/s10994-023-06314-z)  \nUnbiasedNets: A Dataset Diversi􀀌cation Framework for Robustness Bias Alleviation in Neural Networks  \nMahum Naseer 1*, Bharath Srinivas Prabakaran 1 , Osman Hasan2 and Muhammad Sha􀀌que3  \n1* Technische Universit􀁿at Wien (TU Wien), Vienna, 1040, Austria.  \n2 School of Electrical Engineering & Computer Science (SEECS),  \nNational University of Sciences & Technology (NUST), Sector H-12, Islamabad, 44000, Pakistan.  \n3 Division of Engineering, New York University Abu Dhabi  \n(NYUAD), Abu Dhabi, United Arab Emirates.  \n*Corresponding [author E-mail: mahum.naseer@tuwien.ac.at](author E-mail: mahum.naseer@tuwien.ac.at); Contributing authors: [bharath.prabakaran@tuwien.ac.at](bharath.prabakaran@tuwien.ac.at) ; [osman.hasan@seecs.nust.edu.pk](osman.hasan@seecs.nust.edu.pk) ; muhammad.sha􀀌que@nyu.edu;  \nAbstract  \nPerformance of trained neural network (NN) models, in terms of testing accuracy, has improved remarkably over the past several years, especially with the advent of deep learning. However, even the most accurate NNscan be biased toward a speci􀀌c output classi􀀌cation due to the inherent bias in the available training datasets, which may propagate to the real-world implementations. This paper deals with the robustness bias, i.e., the bias exhibited by the trained NN by having a signi􀀌cantly large robustness to noise for a certain output class, as compared to the remaining output classes. The bias is shown to result from imbalanced datasets, i.e., the datasets where all output classes are not equally represented. Towards this, we propose the UnbiasedNets framework, which leverages K-means clustering and the NN's noise tolerance to diversify the given training dataset, even from relatively smaller datasets. This generates balanced datasets and reduces the bias within the datasets themselves. To the best of our knowledge, this is the 􀀌rst framework catering to the robustness bias problem in NNs. We use realworld datasets to demonstrate the e􀀎cacy of the UnbiasedNets for  \ndata diversi􀀌cation, in case of both binary and multi-label classi􀀌ers.  \nThe results are compared to well-known tools aimed at generating balanced datasets, and illustrate how existing works have limited success while addressing the robustness bias. In contrast, UnbiasedNets provides a notable improvement over existing works, while even reducing the robustness bias signi􀀌cantly in some cases, as observed by comparing the NNs trained on the diversi􀀌ed and original datasets.  \nKeywords: Bias, Data-centric bias alleviation, K-means clustering, Neural networks, Noise tolerance  \n1 Introduction  \nMachine learning (ML)-based systems are becoming increasingly ubiquitous in today's world, with their applications ranging from small embedded devices (like health monitoring in smartwatches (Esteva et al. , 2019)) to large safetycritical systems (like autonomous driving (Fink, Liu, Engstle, & Schneider, 2019)) . Their success is often attributed to the Neural Networks (NNs) deployed in these systems, which have the ability to learn and perform decisionmaking with a high accuracy, without being explicitly programmed for their designated task. Typically, these NNs are trained on large datasets, with tens to hundreds of thousands of input samples, using various supervised training algorithms. Testing accuracy is often the most commonly (and possibly the only) used metric to analyze the performance of these NNs.  \nThis spotlights two major limitations: (a) there is a notable reliance on large, labeled datasets, obtaining which is a signi􀀌cant challenge for the ML community, especially for new use-cases, and (b) the trained NN may experience problems like robustness bias, i.e., the ro","cbCaisCFXeJnNoAY","https://ap.wps.com/l/cbCaisCFXeJnNoAY","pdf",2156843,1,33,"English","en",105,"# Introduction\n## Motivating Example","[{\"question\":\"什么是鲁棒性偏置（robustness bias）？\",\"answer\":\"鲁棒性偏置指的是：训练后的神经网络在噪声条件下，对某些输出类别表现出显著更强的鲁棒性，而对其余类别则相对较弱。\"},{\"question\":\"鲁棒性偏置主要由什么原因引起？\",\"answer\":\"文中指出，鲁棒性偏置来源于训练数据集的不均衡（imbalanced datasets），即各输出类别在训练集中并未得到同等数量的表示。\"},{\"question\":\"UnbiasedNets如何缓解鲁棒性偏置？\",\"answer\":\"UnbiasedNets利用K-means聚类与神经网络的噪声容忍能力，对给定训练数据进行多样化处理，从而生成更平衡的数据集并降低数据内部的偏置。\"},{\"question\":\"实验结果表明UnbiasedNets相较现有方法有何改进？\",\"answer\":\"实验中，将UnbiasedNets与面向生成平衡数据集的知名工具进行对比。结果显示现有工作在处理鲁棒性偏置时成功有限，而UnbiasedNets在多种情况下能显著降低鲁棒性偏置并带来明显提升。\"}]","UnbiasedNets - A Dataset Diversification Framework for Robustness Bias Alleviation in Neural Networks | PDF",1785805861,83,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":90,"head_meta":92,"extra_data":94,"updated_unix":28},"unbiasednets-a-dataset-diversification-framework-for-robustness-bias-alleviation-in-neural-networks","",{"@graph":36,"@context":89},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/unbiasednets-a-dataset-diversification-framework-for-robustness-bias-alleviation-in-neural-networks/121637/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81,85],{"name":72,"@type":73,"acceptedAnswer":74},"什么是鲁棒性偏置（robustness bias）？","Question",{"text":75,"@type":76},"鲁棒性偏置指的是：训练后的神经网络在噪声条件下，对某些输出类别表现出显著更强的鲁棒性，而对其余类别则相对较弱。","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"鲁棒性偏置主要由什么原因引起？",{"text":80,"@type":76},"文中指出，鲁棒性偏置来源于训练数据集的不均衡（imbalanced datasets），即各输出类别在训练集中并未得到同等数量的表示。",{"name":82,"@type":73,"acceptedAnswer":83},"UnbiasedNets如何缓解鲁棒性偏置？",{"text":84,"@type":76},"UnbiasedNets利用K-means聚类与神经网络的噪声容忍能力，对给定训练数据进行多样化处理，从而生成更平衡的数据集并降低数据内部的偏置。",{"name":86,"@type":73,"acceptedAnswer":87},"实验结果表明UnbiasedNets相较现有方法有何改进？",{"text":88,"@type":76},"实验中，将UnbiasedNets与面向生成平衡数据集的知名工具进行对比。结果显示现有工作在处理鲁棒性偏置时成功有限，而UnbiasedNets在多种情况下能显著降低鲁棒性偏置并带来明显提升。","https://schema.org",{"og:url":52,"og:type":91,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":93,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":96},[97,101,105,109,114,119,124,127,132,135,139],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Exam",70,"exam",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},5,"Comic",60,"comic",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},6,"Technology",50,"technology",{"id":120,"doc_module":4,"doc_module_name":46,"category_name":121,"show_sort_weight":122,"slug":123},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":125,"slug":126},30,"research-report",{"id":128,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":130,"slug":131},9,"Religion & Spirituality",20,"religion-spirituality",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":130,"slug":134},"World Cup","world-cup",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":136,"slug":138},10,"Lifestyle","lifestyle",{"id":140,"doc_module":4,"doc_module_name":46,"category_name":141,"show_sort_weight":110,"slug":142},19,"General","general"]