Description
Deep neural networks (DNNs) undergo lengthy and expensive training procedures whose outcome - the DNN weights - represents a significant intellectual property asset to protect.
Side-channel analysis (SCA) has recently appeared as an effective approach to recover this confidential asset of DNN implementations.
Ding et al. (HOST’25) introduced MACPRUNING, a novel SCA countermeasure based on pruning, a performance-oriented Approximate Computing technique: at inference time, the implementation randomly prunes (or skips) non-important weights (i.e., with low contribution to the DNN’s accuracy) of the first layer, exponentially increasing the side-channel resilience of the protected DNN implementation.
In this presentation, we describe a preprocessing methodology taking advantage of a control-flow dependency intrinsic to the countermeasure's design.
Through practical experiments, we demonstrate the effectiveness of our methodology, recovering up to 96% of the important weights of a MACPruning-protected Multi-Layer Perceptron.
Moreover, we show how microarchitectural leakage improves the effectiveness of our methodology, even allowing for the recovery of up to 100% of the targeted non-important weights.
Lastly, by adapting our methodology, we elaborate on how the pruning mechanism, which depends on the importance of the weights, enables the circumvention of a control-flow-free MACPRUNING implementation.
With this last point, we identify the pruning mechanism underlying MACPRUNING as the root of the countermeasure’s vulnerability.
Autre
Présentation en Anglais (slides en Anglais)
Talk in English (slides in English)
Practical infos
Next sessions
-
An Assessment of the Security of Reconfigurable Cloud Computing
Speaker : Russell Tessier - University of Massachusetts
In this talk, we examine efforts to protect network processing and computation in cloud-based FPGAs from hardware- and software-based attacks. FPGAs have served as integral components in cloud computing for a decade, even though commercial FPGA architectures were not designed for cloud deployment. A range of potential attacks on cloud-based FPGAs and their remediations are examined. We[…]-
SemSecuElec
-
-
Overcoming Challenges in Advanced Application Processor Attacks Using PEM-Assisted LFI
Speaker : Jean-Max Dutertre - Ecole Mines Saint Etienne
Modern Application Processors (or SoC) are complex multi-cores devices that run application code through an operating system. Manufactured using advanced FinFET technology, they may feature billions of transistors. They also operate at clock frequencies of over 1 GHz. This complexity is often presented as a barrier against fault injection attacks. Indeed, a potential attacker, would face two main[…]-
SemSecuElec
-
Fault injection
-
-
Securing processor's microarchitecture against SCA in a post-quantum cryptography setting
Speaker : Vincent MIGLIORE - LAAS-CNRS
Hardware microarchitecture is a well-known source of side-channel leakages, providing a notable security reduction of standard cryptographic algorithms (e.g. AES) if not properly addressed by software or hardware. In this talk, we present new design approaches to harden processor's microarchitecture against power-based side-channel attacks, relying on configurable and cascadable building blocks[…]-
SemSecuElec
-
Side-channel
-
Micro-architectural vulnerabilities
-
-
Compositional Semantics for Detection and Defense of Speculative Leaks
Speaker : Xaver Fabian - CISPA Helmholtz Center for Information Security
Speculative execution improves CPU performance but introduces vulnerabilities, known as Spectre attacks, which exploit microarchitectural side effects to leak sensitive data. Modern processors employ multiple interacting speculation mechanisms, yet existing approaches predominantly analyze them in isolation, which limits both the automated detection of complex leaks and the formal verification of[…]-
SemSecuElec
-
Side-channel
-
Micro-architectural vulnerabilities
-