A Simple Transformer Pipeline for Full-Key Side-Channel Attacks on Uncropped Datasets
2026-08-31 • Cryptography and Security
Cryptography and SecurityMachine Learning
AI summaryⓘ
The authors created a straightforward transformer model to analyze side-channel data without chopping it up, enabling attacks on the entire cryptographic key at once. Their approach uses a standard transformer design with minor changes to fit side-channel inputs and outputs. They provide open-source code, training instructions, and pretrained models that work well on popular datasets, requiring modest computing power and time to train. This shows that a simple transformer can be an effective baseline for these security analyses.
Deep LearningSide-Channel AnalysisTransformer ModelFull-Key AttackUncropped TracesASCAD DatasetPretrained WeightsNVIDIA A6000Encoder BackboneCryptography
Authors
Jimmy Gammell, Kaushik Roy
Abstract
Deep learning-based side-channel analysis has historically focused on single-byte targets and manually cropped traces, which risks discarding exploitable leakage. While recent work has proposed specialized architectures and resampling techniques to address this gap, the literature lacks a simple transformer baseline for simultaneous full-key attacks on uncropped traces. We present an open-source transformer implementation for uncropped full-key attacks which uses the standard transformer encoder backbone, adapting only the input and output layers to the side-channel setting. We release our implementation, training recipes, and pretrained weights for uncropped ASCADv1f, ASCADv1r, and CHES-CTF-2018 which achieve performance competitive with previously-reported results, while using less than 10GB of VRAM and requiring at most 3.34 hours of training on a single NVIDIA A6000.