Robustness Verification for Transformers

Abstract

Robustness verification that aims to formally certify the prediction behavior of neural networks has become an important tool for understanding model behavior and obtaining safety guarantees. However, previous methods can usually only handle neural networks with relatively simple architectures. In this paper, we consider the robustness verification problem for Transformers. Transformers have complex self-attention layers that pose many challenges for verification, including cross-nonlinearity and cross-position dependency, which have not been discussed in previous works. We resolve these challenges and develop the first robustness verification algorithm for Transformers. The certified robustness bounds computed by our method are significantly tighter than those by naive Interval Bound Propagation. These bounds also shed light on interpreting Transformers as they consistently reflect the importance of different words in sentiment analysis.

arXiv commentICLR 2020

References in corpus (20)

• XLNet: Generalized Autoregressive Pretraining for Language Understanding
• VisualBERT: A Simple and Performant Baseline for Vision and Language
• VL-BERT: Pre-training of Generic Visual-Linguistic Representations
• Provable defenses against adversarial examples via the convex outer adversarial polytope
• Generating Natural Adversarial Examples
• Efficient Neural Network Robustness Certification with General Activation Functions
• Evaluating Robustness of Neural Networks with Mixed Integer Programming
• Formal Security Analysis of Neural Networks using Symbolic Intervals
• On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Models
• Formal Guarantees on the Robustness of a Classifier against Adversarial Manipulation
• Semidefinite relaxations for certifying robustness to adversarial examples
• Efficient Formal Safety Analysis of Neural Networks
• A Dual Approach to Scalable Verification of Deep Networks
• A Unified View of Piecewise Linear Neural Network Verification
• POPQORN: Quantifying Robustness of Recurrent Neural Networks
• Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation
• Verifying Aircraft Collision Avoidance Neural Networks Through Linear Approximations of Safe Regions
• Verification of Recurrent Neural Networks Through Rule Extraction
• Correctness Verification of Neural Networks
• Verification of Non-Linear Specifications for Neural Networks

Cited by in corpus (14)

• Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond
• Natural Language Adversarial Defense through Synonym Encoding
• Defense against Adversarial Attacks in NLP via Dirichlet Neighborhood Ensemble
• Towards a Robust Deep Neural Network in Texts: A Survey
• Robust Encodings: A Framework for Combating Adversarial Typos
• Improving the Robustness of Transformer-based Large Language Models with Dynamic Attention
• Improving Robustness and Generality of NLP Models Using Disentangled Representations
• Learning Neural Models for Natural Language Processing in the Face of Distributional Shift
• Improving Zero-Shot Cross-Lingual Transfer Learning via Robust Training
• Certified Robustness to Programmable Transformations in LSTMs
• Achieving Model Robustness through Discrete Adversarial Training
• Searching for an Effective Defender: Benchmarking Defense against Adversarial Word Substitution
• What's in a Name? Are BERT Named Entity Representations just as Good for any other Name?
• Robustness to Programmable String Transformations via Augmented Abstract Training