Formal Security Analysis of Neural Networks using Symbolic Intervals
This addresses the need for efficient formal security guarantees in DNNs deployed in critical domains like autonomous vehicles, representing a novel direction rather than an incremental improvement.
The paper tackles the problem of formally verifying security properties of deep neural networks (DNNs) by introducing a method based on symbolic interval analysis instead of SMT solvers, resulting in a system called ReluVal that outperforms the state-of-the-art solver-based system by 200 times on average and verifies properties that previously timed out.
Due to the increasing deployment of Deep Neural Networks (DNNs) in real-world security-critical domains including autonomous vehicles and collision avoidance systems, formally checking security properties of DNNs, especially under different attacker capabilities, is becoming crucial. Most existing security testing techniques for DNNs try to find adversarial examples without providing any formal security guarantees about the non-existence of such adversarial examples. Recently, several projects have used different types of Satisfiability Modulo Theory (SMT) solvers to formally check security properties of DNNs. However, all of these approaches are limited by the high overhead caused by the solver. In this paper, we present a new direction for formally checking security properties of DNNs without using SMT solvers. Instead, we leverage interval arithmetic to compute rigorous bounds on the DNN outputs. Our approach, unlike existing solver-based approaches, is easily parallelizable. We further present symbolic interval analysis along with several other optimizations to minimize overestimations of output bounds. We design, implement, and evaluate our approach as part of ReluVal, a system for formally checking security properties of Relu-based DNNs. Our extensive empirical results show that ReluVal outperforms Reluplex, a state-of-the-art solver-based system, by 200 times on average. On a single 8-core machine without GPUs, within 4 hours, ReluVal is able to verify a security property that Reluplex deemed inconclusive due to timeout after running for more than 5 days. Our experiments demonstrate that symbolic interval analysis is a promising new direction towards rigorously analyzing different security properties of DNNs.