https://github.com/abdulfatir/redsds

Pytorch implementation of RED-SDS (NeurIPS 2021).
https://github.com/abdulfatir/redsds

Last synced: 6 months ago
JSON representation

Pytorch implementation of RED-SDS (NeurIPS 2021).

• Host: GitHub
• URL: https://github.com/abdulfatir/redsds
• Owner: abdulfatir
• Created: 2021-10-26T13:33:23.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2021-10-27T05:43:25.000Z (almost 4 years ago)
• Last Synced: 2025-03-31T00:41:17.671Z (7 months ago)
• Language: Python
• Homepage:
• Size: 59.6 KB
• Stars: 18
• Watchers: 2
• Forks: 5
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Recurrent Explicit Duration Switching Dynamical Systems (RED-SDS)

[![ArXiv](https://img.shields.io/static/v1?label=arXiv&message=2110.13878&color=B31B1B)](https://arxiv.org/abs/2110.13878)

[![Venue:NeurIPS 2021](https://img.shields.io/badge/Venue-NeurIPS%202021-007CFF)](https://neurips.cc)

This repository contains a reference implementation of RED-SDS, a non-linear state space model proposed in the NeurIPS 2021 paper [*Deep Explicit Duration Switching Models for Time Series*](https://arxiv.org/abs/2110.13878). 

## Environment Setup

* Run `pip install -r requirements.txt`.

## Usage

### Reevaluating Trained Models

* Download the trained models from [this link](https://drive.google.com/drive/folders/19DmHwmsDZGbU4WEujwp0j1HWMdA5dDhU?usp=sharing).

* Run `python reevaluate.py --ckpt .pt`.

### Training Models

#### Segmentation

* Generate/download datasets.

    * To generate the bouncing ball and 3 mode system datasets, use the notebooks in `./data/`. Alternatively, you can download the datasets from [this link](https://drive.google.com/drive/folders/1g5O2jktqWnH2p1BCkn1WtBBmSEF5AMTD?usp=sharing).

    * To download and preprocess the dancing bees dataset, run `./data/bee.sh`.

* Run `python run_segmentation.py --config configs/.yaml --device cuda:0` to train the RED-SDS model.

* Run `tensorboard --logdir /path/to/results/dir` to visualize results.

#### Forecasting

* Run `python run_gts_univariate.py --config configs/.yaml --device cuda:0` to train the RED-SDS model. The dataset will be downloaded automatically.

* Run `tensorboard --logdir /path/to/results/dir` to visualize results.

## Questions

For any questions regarding the code or the paper, please email [Fatir](mailto:abdulfatir@u.nus.edu), [Konstantinos](mailto:kbenidis@amazon.com), or [Richard](mailto:kurler@amazon.com).

## BibTeX

If you find this repository or the ideas presented in our paper useful for your research, please consider citing our paper.

```

@inproceedings{ansari2021deep,

  author    = {Abdul Fatir Ansari and Konstantinos Benidis and Richard Kurle and Ali Caner Turkmen and Harold Soh and Alex Smola and Bernie Wang and Tim Januschowski},

  title     = {Deep Explicit Duration Switching Models for Time Series},

  year      = {2021},

  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},

}

```

## Acknowledgement

This repo contains parts of code based on the following repos:

| Repo  | Copyright (c) | License |

| ------------- | ---------- | ------------- |

| [google-research/google-research/snlds](https://github.com/google-research/google-research/tree/master/snlds)  | The Google Research Authors | [Apache 2.0](http://www.apache.org/licenses/LICENSE-2.0) |

| [mattjj/pyslds](https://github.com/mattjj/pyslds) | Matthew James Johnson | [MIT](https://github.com/mattjj/pyslds/blob/master/LICENSE-MIT)