Semi-Supervised Sequence Classification through Change Point Detection
This repository contains code for "Semi-supervised sequence classification through Change Point Detection", AAAI 2021
Requirements
- Python 3.7+
- Numpy
- Pytorch
- MATLAB 2018+
The main.py script is used to train neural networks to obtain representations through sim/dissim pairs from change point detection.
The detect_cp folder contains the script run_cp.py which is used to detect change points. The script outputs similar dissimilar pairs which are used by main.py
Datafiles for this code can be donwloaded from:
https://www.dropbox.com/s/a6035sz2wdowkas/semi_sup_cp_datafiles.zip?dl=0