121 citations · 295 across the 15 of their papers we have counts for
15 papers
Provable Defense against Backdoor Policies in Reinforcement Learning
Shubham Kumar Bharti, Xuezhou Zhang, Adish Singla +1
We propose a provable defense mechanism against backdoor policies in reinforcement learning under subspace trigger assumption. A backdoor policy is a security threat where an adver…
Controllable and Diverse Text Generation in E-commerce
Huajie Shao, Jun Wang, Haohong Lin +4
In E-commerce, a key challenge in text generation is to find a good trade-off between word diversity and accuracy (relevance) in order to make generated text appear more natural an…
Reward Poisoning in Reinforcement Learning: Attacks Against Unknown Learners in Unknown Environments
Amin Rakhsha, Xuezhou Zhang, Xiaojin Zhu +1
We study black-box reward poisoning attacks against reinforcement learning (RL), in which an adversary aims to manipulate the rewards to mislead a sequence of RL agents with unknow…
Robust Policy Gradient against Strong Data Corruption
Xuezhou Zhang, Yiding Chen, Xiaojin Zhu +1
We study the problem of robust reinforcement learning under adversarial corruption on both rewards and transitions. Our attack model assumes an \textit{adaptive} adversary who can…
Task-agnostic Exploration in Reinforcement Learning
Xuezhou Zhang, Yuzhe ma, Adish Singla
Efficient exploration is one of the main challenges in reinforcement learning (RL). Most existing sample-efficient algorithms assume the existence of a single reward function durin…
The Sample Complexity of Teaching-by-Reinforcement on Q-Learning
Xuezhou Zhang, Shubham Kumar Bharti, Yuzhe Ma +2
We study the sample complexity of teaching, termed as "teaching dimension" (TDim) in the literature, for the teaching-by-reinforcement paradigm, where the teacher guides the studen…
Neural Additive Models: Interpretable Machine Learning with Neural Nets
Rishabh Agarwal, Levi Melnick, Nicholas Frosst +4
Deep neural networks (DNNs) are powerful black-box predictors that have achieved impressive performance on a wide variety of tasks. However, their accuracy comes at the cost of int…
Adaptive Reward-Poisoning Attacks against Reinforcement Learning
Xuezhou Zhang, Yuzhe Ma, Adish Singla +1
In reward-poisoning attacks against reinforcement learning (RL), an attacker can perturb the environment reward into at each step, with the goal of forcing the RL a…
Policy Poisoning in Batch Reinforcement Learning and Control
Yuzhe Ma, Xuezhou Zhang, Wen Sun +1
We study a security threat to batch reinforcement learning and control where the attacker aims to poison the learned policy. The victim is a reinforcement learner / controller whic…
Online Data Poisoning Attack
Xuezhou Zhang, Xiaojin Zhu, Laurent Lessard
We study data poisoning attacks in the online setting where training items arrive sequentially, and the attacker may perturb the current item to manipulate online learning. Importa…
Training Set Camouflage
Ayon Sen, Scott Alfeld, Xuezhou Zhang +3
We introduce a form of steganography in the domain of machine learning which we call training set camouflage. Imagine Alice has a training set on an illicit machine learning classi…
An Optimal Control Approach to Sequential Machine Teaching
Laurent Lessard, Xuezhou Zhang, Xiaojin Zhu
Given a sequential learning algorithm and a target model, sequential machine teaching aims to find the shortest training sequence to drive the learning algorithm to the target mode…
Axiomatic Interpretability for Multiclass Additive Models
Xuezhou Zhang, Sarah Tan, Paul Koch +3
Generalized additive models (GAMs) are favored in many regression and binary classification problems because they are able to fit complex, nonlinear functions while still remaining…
Teacher Improves Learning by Selecting a Training Subset
Yuzhe Ma, Robert Nowak, Philippe Rigollet +2
We call a learner super-teachable if a teacher can trim down an iid training set while making the learner learn even better. We provide sharp super-teaching guarantees on two learn…
Training Set Debugging Using Trusted Items
Xuezhou Zhang, Xiaojin Zhu, Stephen J. Wright
Training set bugs are flaws in the data that adversely affect machine learning. The training set is usually too large for man- ual inspection, but one may have the resources to ver…