Exploiting Representation Curvature for Boundary Detection in Time Series

Publication
Shin, Y., Park, J., Yoon, S., Song, H., Lee, B., Lee, J., "Exploiting Representation Curvature for Boundary Detection in Time Series", In Proceedings of Conference on Neural Information Processing Systems (NeurIPS), 2024. [link]

This repository is the official PyTorch implementation of RECURVE.

How to Install

1. We require following packages to run the code. Please download all the requirements in your python environment.
   • python 3.9.15
   • pytorch 1.13.1
   • numpy 1.25.0
   • pandas 1.4.4
   • cuda 11.7.1
   • scipy 1.11.1
   • scikit-learn 1.2.2

Dataset

Datasets are in /dataset and should be preprocessed first using preprocessing.ipynb. After preprocessing, datasets are converted into .npy format in /dataset. .npy files of {HAPT, mHealth, WISDM} is available in the repository. For 50salads dataset, please download the dataset in this url.

How to Run

At current directory which has all source codes, run main.py to get AUC and LOC score of RECURVE.

• dataset: {mHealth, HAPT, WISDM, 50salads} # designate which dataset to use.
• seed: {0, 1, 2, 3, 4} # seed for 5-fold cross validation.
• gpu: an integer for gpu id
• repr: {TSCP2, TNC} # representing TPC and TNC representation learning methods e.g.) python3 main.py --data HAPT --repr TSCP2 --gpu 0 --seed 0