Adversarial Reinforced Instruction Attacker for Robust Vision-Language Navigation

Abstract

Language instruction plays an essential role in the natural language grounded navigation tasks. However, navigators trained with limited human-annotated instructions may have difficulties in accurately capturing key information from the complicated instruction at different timesteps, leading to poor navigation performance. In this paper, we exploit to train a more robust navigator which is capable of dynamically extracting crucial factors from the long instruction, by using an adversarial attacking paradigm. Specifically, we propose a Dynamic Reinforced Instruction Attacker (DR-Attacker), which learns to mislead the navigator to move to the wrong target by destroying the most instructive information in instructions at different timesteps. By formulating the perturbation generation as a Markov Decision Process, DR-Attacker is optimized by the reinforcement learning algorithm to generate perturbed instructions sequentially during the navigation, according to a learnable attack score. Then, the perturbed instructions, which serve as hard samples, are used for improving the robustness of the navigator with an effective adversarial training strategy and an auxiliary self-supervised reasoning task. Experimental results on both Vision-and-Language Navigation (VLN) and Navigation from Dialog History (NDH) tasks show the superiority of our proposed method over state-of-the-art methods. Moreover, the visualization analysis shows the effectiveness of the proposed DR-Attacker, which can successfully attack crucial information in the instructions at different timesteps. Code is available at https://github.com/expectorlin/DR-Attacker.

arXiv commentAccepted by TPAMI 2021

References in corpus (19)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• VQA: Visual Question Answering
• Robust Adversarial Reinforcement Learning
• FreeLB: Enhanced Adversarial Training for Natural Language Understanding
• Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers
• Building Generalizable Agents with a Realistic and Rich 3D Environment
• Speaker-Follower Models for Vision-and-Language Navigation
• Word-level Textual Adversarial Attacking as Combinatorial Optimization
• On Adversarial Examples for Character-Level Neural Machine Translation
• Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
• Self-Monitoring Navigation Agent via Auxiliary Progress Estimation
• Defense Against Adversarial Attacks Using Feature Scattering-based Adversarial Training
• Vision-and-Dialog Navigation
• Adversarial Training for Large Neural Language Models
• InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective
• Improving Neural Language Modeling via Adversarial Training
• Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems
• The Regretful Agent: Heuristic-Aided Navigation through Progress Estimation
• On Adversarial Examples for Biomedical NLP Tasks

Cited by in corpus (2)

• PASTS: Progress-Aware Spatio-Temporal Transformer Speaker For Vision-and-Language Navigation
• Multimodal Transformer with Variable-length Memory for Vision-and-Language Navigation