权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

III: Small: Usable Interpretability

III：小：可用的可解释性

基本信息

批准号：
1951729
负责人：
Ting Wang
金额：
$ 49.56万
依托单位：
Pennsylvania State Univ University Park
依托单位国家：
美国
项目类别：
Continuing Grant
财政年份：
2019
资助国家：
美国
起止时间：
2019-08-26 至 2024-03-31
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=1951729&HistoricalAwards=false
关键词：
III Small Usable Interpretability

项目摘要

Deep neural network (DNN)-powered systems and services hold great promise to fundamentally transform the way people live, work and play. Yet, to fully unleash this potential, it is critical to improve their interpretability to make them more trustworthy and easy-to-use. The transformative nature of this project is to completely rethink how to define and implement the interpretation of DNNs and how to exploit this interpretability as a bridge to understand and control the DNN behaviors. The success of this project will not only improve the reliability, interactivity and operability of DNN-powered systems, but also promote more principled practice of building and using machine learning systems in general. The research products will be applicable to fields including machine learning, cyber-security and human-computer interaction.This project aims to develop RIDDLE, a new interpretable deep learning framework that is reliable, because it deploys built-in defenses against adversarial manipulations; interactive, because it provides interfaces and mechanisms to perform in-depth, interactive analysis of DNN dynamics; and debuggable, because it employs interpretability as the lens for users to effectively control DNN behaviors. Along the three directions, the specific tasks of this project include: exploring the vulnerabilities of existing interpretation models to adversarial manipulations, uncovering their root causes, and developing practical defense mechanisms, designing an expressive interpretation algebra framework to allow users to flexibly construct interactive analysis tools for a variety of DNNs and tasks, which circumvents the "one-size-fits-all" challenge, and building interpretation-based model debugging techniques, which allow users to effectively localize and fix model defects.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

深度神经网络（DNN）驱动的系统和服务有望从根本上改变人们的生活、工作和娱乐方式。然而，要充分释放这种潜力，关键是要提高它们的可解释性，使它们更值得信赖和易于使用。该项目的变革性质是彻底重新思考如何定义和实现深度神经网络的解释，以及如何利用这种可解释性作为理解和控制深度神经网络行为的桥梁。该项目的成功不仅将提高dnn驱动系统的可靠性、交互性和可操作性，而且还将促进构建和使用机器学习系统的更有原则的实践。研究成果将适用于机器学习、网络安全和人机交互等领域。该项目旨在开发RIDDLE，这是一种新的可解释深度学习框架，它是可靠的，因为它部署了针对对抗性操作的内置防御；交互式，因为它提供了接口和机制来执行深度神经网络动态的交互式分析；并且是可调试的，因为它采用可解释性作为透镜，让用户有效地控制DNN的行为。沿着这三个方向，本项目的具体任务包括：探索现有解释模型对对抗性操作的脆弱性，揭示其根本原因，并开发实用的防御机制，设计一个富有表现力的解释代数框架，允许用户灵活地为各种dnn和任务构建交互式分析工具，从而规避“一刀切”的挑战，并构建基于解释的模型调试技术。它允许用户有效地定位和修复模型缺陷。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

期刊论文数量（36）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

Transfer Attacks Revisited: A Large-Scale Empirical Study in Real Computer Vision Settings

重新审视传输攻击：真实计算机视觉设置中的大规模实证研究

DOI：
10.1109/sp46214.2022.9833783
发表时间：
2022
期刊：
2022 IEEE Symposium on Security and Privacy
影响因子：
0
作者：
Mao, Yuhao;Fu, Chong;Wang, Saizhuo;Ji, Shouling;Zhang, Xuhong;Liu, Zhenguang;Zhou, Jun;Liu, Alex X.;Beyah, Raheem;Wang, Ting
通讯作者：
Wang, Ting

FOAP: Fine-Grained Open-World Android App Fingerprinting

DOI：
发表时间：
2022
期刊：
影响因子：
0
作者：
Jianfeng Li;Hao Zhou;Shuohan Wu;Xiapu Luo;Ting Wang;Xian Zhan;Xiaobo Ma
通讯作者：
Jianfeng Li;Hao Zhou;Shuohan Wu;Xiapu Luo;Ting Wang;Xian Zhan;Xiaobo Ma

Adversarial CAPTCHAs

对抗性验证码

DOI：
10.1109/tcyb.2021.3071395
发表时间：
2019-01
期刊：
{IEEE} Trans. Cybern.
影响因子：
0
作者：
Chenghui Shi;Xiaogang Xu;Shouling Ji;Kai Bu;Jianhai Chen;Raheem Beyah;Ting Wang
通讯作者：
Ting Wang

Too Good to Be Safe: Tricking Lane Detection in Autonomous Driving with Crafted Perturbations

DOI：
发表时间：
2021
期刊：
影响因子：
0
作者：
Pengfei Jing;Qiyi Tang;Yue Du;Lei Xue;Xiapu Luo;Ting Wang;Sen Nie;Shi Wu
通讯作者：
Pengfei Jing;Qiyi Tang;Yue Du;Lei Xue;Xiapu Luo;Ting Wang;Sen Nie;Shi Wu

AutoML in The Wild: Obstacles, Workarounds, and Expectations

DOI：
10.1145/3544548.3581082
发表时间：
2023-02
期刊：
Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems
影响因子：
0
作者：
Yuan Sun;Qiurong Song;Xinning Gui;Fenglong Ma;Ting Wang
通讯作者：
Yuan Sun;Qiurong Song;Xinning Gui;Fenglong Ma;Ting Wang