NeuroUnlock: Unlocking the Architecture of Obfuscated Deep Neural Networks

Abstract

The advancements of deep neural networks (DNNs) have led to their deployment in diverse settings, including safety and security-critical applications. As a result, the characteristics of these models (e.g., the architecture of layers and weight values/distributions) have become sensitive intellectual properties that require protection from malicious users. Extracting the architecture of a DNN through leaky side-channels (e.g., memory access) allows adversaries to (i) clone the model (i.e., build proxy models with similar accuracy profiles), and (ii) craft adversarial attacks. DNN obfuscation thwarts side-channel-based architecture stealing (SCAS) attacks by altering the run-time traces of a given DNN while preserving its functionality.In this work, we expose the vulnerability of state-of-the-art DNN obfuscation methods (based on predictable and reversible modifications employed in a given DNN architecture) to these attacks. We present NeuroUnlock, a novel SCAS attack against obfuscated DNNs. Our NeuroUnlock employs a sequence-tosequence model that learns the obfuscation procedure and automatically reverts it, thereby recovering the original DNN architecture. We demonstrate the effectiveness of NeuroUnlock by recovering the architecture of 200 randomly generated and obfuscated DNNs running on the Nvidia RTX 2080 TI graphics processing unit (GPU). Moreover, NeuroUnlock recovers the architecture of various other obfuscated (and publicly available) DNNs, such as the VGG-11, VGG-13, ResNet-20, and ResNet-32 networks. After recovering the architecture, NeuroUnlock automatically builds a near-equivalent DNN with only a 1.4% drop in the testing accuracy. We further show that launching a subsequent adversarial attack on the recovered DNNs boosts the success rate of the adversarial attack by 51.7% in average compared to launching it on the obfuscated versions. Additionally, we propose a novel methodology for DNN obfuscation, ReDLock, which eradicates the deterministic nature of the obfuscation and achieves 2.16× more resilience to the NeuroUnlock attack. We release the NeuroUnlock and the ReDLock as open-source frameworks 1 .