BDD4BNN: A BDD-based Quantitative Analysis Framework for Binarized Neural Networks
arXiv:2103.07224
Abstract
Verifying and explaining the behavior of neural networks is becoming increasingly important, especially when they are deployed in safety-critical applications. In this paper, we study verification problems for Binarized Neural Networks (BNNs), the 1-bit quantization of general real-numbered neural networks. Our approach is to encode BNNs into Binary Decision Diagrams (BDDs), which is done by exploiting the internal structure of the BNNs. In particular, we translate the input-output relation of blocks in BNNs to cardinality constraints which are then encoded by BDDs. Based on the encoding, we develop a quantitative verification framework for BNNs where precise and comprehensive analysis of BNNs can be performed. We demonstrate the application of our framework by providing quantitative robustness analysis and interpretability for BNNs. We implement a prototype tool BDD4BNN and carry out extensive experiments which confirm the effectiveness and efficiency of our approach.
References in corpus (9)
- Binarized Neural Networks
- Provable defenses against adversarial examples via the convex outer adversarial polytope
- Interpretable Machine Learning -- A Brief History, State-of-the-Art and Challenges
- Evaluating Robustness of Neural Networks with Mixed Integer Programming
- Distilling a Neural Network Into a Soft Decision Tree
- An approach to reachability analysis for feed-forward ReLU neural networks
- A Game-Based Approximate Verification of Deep Neural Networks with Provable Guarantees
- Optimization and Abstraction: A Synergistic Approach for Analyzing Neural Network Robustness
- Scalable Verification of Quantized Neural Networks (Technical Report)