利用蒸馏辅助可学习方法实现无梯度的事后可解释性

Debarpan Bhattacharya, Amir H. Poorjam, Deepak Mittal, Sriram Ganapathy
{"title":"利用蒸馏辅助可学习方法实现无梯度的事后可解释性","authors":"Debarpan Bhattacharya, Amir H. Poorjam, Deepak Mittal, Sriram Ganapathy","doi":"arxiv-2409.11123","DOIUrl":null,"url":null,"abstract":"The recent advancements in artificial intelligence (AI), with the release of\nseveral large models having only query access, make a strong case for\nexplainability of deep models in a post-hoc gradient free manner. In this\npaper, we propose a framework, named distillation aided explainability (DAX),\nthat attempts to generate a saliency-based explanation in a model agnostic\ngradient free application. The DAX approach poses the problem of explanation in\na learnable setting with a mask generation network and a distillation network.\nThe mask generation network learns to generate the multiplier mask that finds\nthe salient regions of the input, while the student distillation network aims\nto approximate the local behavior of the black-box model. We propose a joint\noptimization of the two networks in the DAX framework using the locally\nperturbed input samples, with the targets derived from input-output access to\nthe black-box model. We extensively evaluate DAX across different modalities\n(image and audio), in a classification setting, using a diverse set of\nevaluations (intersection over union with ground truth, deletion based and\nsubjective human evaluation based measures) and benchmark it with respect to\n$9$ different methods. In these evaluations, the DAX significantly outperforms\nthe existing approaches on all modalities and evaluation metrics.","PeriodicalId":501289,"journal":{"name":"arXiv - EE - Image and Video Processing","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Gradient-free Post-hoc Explainability Using Distillation Aided Learnable Approach\",\"authors\":\"Debarpan Bhattacharya, Amir H. Poorjam, Deepak Mittal, Sriram Ganapathy\",\"doi\":\"arxiv-2409.11123\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The recent advancements in artificial intelligence (AI), with the release of\\nseveral large models having only query access, make a strong case for\\nexplainability of deep models in a post-hoc gradient free manner. In this\\npaper, we propose a framework, named distillation aided explainability (DAX),\\nthat attempts to generate a saliency-based explanation in a model agnostic\\ngradient free application. The DAX approach poses the problem of explanation in\\na learnable setting with a mask generation network and a distillation network.\\nThe mask generation network learns to generate the multiplier mask that finds\\nthe salient regions of the input, while the student distillation network aims\\nto approximate the local behavior of the black-box model. We propose a joint\\noptimization of the two networks in the DAX framework using the locally\\nperturbed input samples, with the targets derived from input-output access to\\nthe black-box model. We extensively evaluate DAX across different modalities\\n(image and audio), in a classification setting, using a diverse set of\\nevaluations (intersection over union with ground truth, deletion based and\\nsubjective human evaluation based measures) and benchmark it with respect to\\n$9$ different methods. In these evaluations, the DAX significantly outperforms\\nthe existing approaches on all modalities and evaluation metrics.\",\"PeriodicalId\":501289,\"journal\":{\"name\":\"arXiv - EE - Image and Video Processing\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - EE - Image and Video Processing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.11123\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - EE - Image and Video Processing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.11123","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

最近,人工智能(AI)取得了长足的进步,发布了多个大型模型,这些模型只允许查询访问,这有力地证明了深度模型可以通过无后置梯度的方式进行解释。在本文中,我们提出了一个名为 "蒸馏辅助可解释性(DAX)"的框架,该框架试图在模型不可知论的无梯度应用中生成基于显著性的解释。DAX 方法通过一个掩码生成网络和一个蒸馏网络,在可学习的环境中提出了解释问题。掩码生成网络通过学习来生成乘法掩码,从而找到输入的突出区域,而学生蒸馏网络则旨在逼近黑盒模型的局部行为。我们提出了在 DAX 框架中使用局部扰动输入样本对这两个网络进行联合优化的方法,而目标则来自黑盒模型的输入输出访问。我们在不同模式(图像和音频)的分类环境中广泛评估了 DAX,使用了一系列不同的评估方法(与地面实况的交集与联合、基于删除的评估方法和基于主观人类评估的评估方法),并将其与 9 种不同的方法进行比较。在这些评估中,DAX 在所有模式和评估指标上都明显优于现有方法。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Gradient-free Post-hoc Explainability Using Distillation Aided Learnable Approach
The recent advancements in artificial intelligence (AI), with the release of several large models having only query access, make a strong case for explainability of deep models in a post-hoc gradient free manner. In this paper, we propose a framework, named distillation aided explainability (DAX), that attempts to generate a saliency-based explanation in a model agnostic gradient free application. The DAX approach poses the problem of explanation in a learnable setting with a mask generation network and a distillation network. The mask generation network learns to generate the multiplier mask that finds the salient regions of the input, while the student distillation network aims to approximate the local behavior of the black-box model. We propose a joint optimization of the two networks in the DAX framework using the locally perturbed input samples, with the targets derived from input-output access to the black-box model. We extensively evaluate DAX across different modalities (image and audio), in a classification setting, using a diverse set of evaluations (intersection over union with ground truth, deletion based and subjective human evaluation based measures) and benchmark it with respect to $9$ different methods. In these evaluations, the DAX significantly outperforms the existing approaches on all modalities and evaluation metrics.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信