This is the code of Behavior Sequence Anomaly Detection Approach The project mainly consists of the following components:
- pre-log: This directory contains the implementation for preprocessing the DroidEvolver dataset.
- shapelet_generation: This directory is used to extract the shapelets of sequences from the input data.
- Lcs-transform: This directory contains the implementation for the long common subsequence (LCS) algorithm to extract behavior units from the perturbed behavior sequence
- em_add__transformer_train.py: This is the main code for implementing Feature Extraction and Behavior Classification
- seq-gan: This directory：The seqgan directory is used for generating sequence sample fragments using Generative Adversarial Networks (GANs).
- defense-gan: The code in this directory contains the implementation of Defense-GAN.