This repository contains the datasets used for deep learning for our paper "Combining Loop Shuffling and Code Polymorphism for Enhanced AES Side-Channel Security", published at COSADE 2024. It contains:- A dataset of traces measured on a STM32F7. Detailed experimental setup is available in the paper. The traces measured are long enough to contain the full first round of the AES.- A dataset of simulated traces. Simulated traces output 2 samples for each instruction executed: one sample consisting of the sum of the Hamming weights of the input registers, and one sample consisting of (the sum of) the Hamming weight(s) of the output register(s). The traces contain all rounds of the AES. Note that the plaintexts file is different in both dataset. **IMPORTANT NOTE:**This dataset stems from a serie of experiments carried on a particular implementation of AES, with particular configurations of the countermeasures. We stress that other security levels (lower or higher) can be reached with the considered countermeasures depending on the platform, the implementation, the configuration of the countermeasures etc. In particular, the code polymorphism countermeasure is highly configurable, and this dataset is limited to the configuration described in the paper.