This is a PyTorch implementation of Series2Vec: Similarity-based Self-supervised Representation Learning for Time Series Classification.

If you downloaded the code prior to the latest update, please ensure to update to the current version as it is consistent with the paper.

The datasets used for training and evaluation can be downloaded from the following locations:

Download the datasets from this Google Drive link.
After downloading, place them in the Datasets/Benchmarks/ directory.

You can download it from the official UEA website.

• Paper: Series2Vec Paper on Springer
• Blog Post: Meet Series2Vec: A New Way to Decode Time Series

Instructions are for Unix-based systems (e.g., Linux, MacOS).

To see all command options with explanations, run: python main.py --help. In utils/args.py you can select the datasets and modify the model parameters. For example:

self.parser.add_argument('--epochs', type=int, default=100, help='Number of training epochs')

or you can set the parameters:

python main.py --dataset Benchmarks

If you find Series2vec useful for your research, please consider citing this paper using the following information:

```
@article{series2vec2024,
  title={Series2vec: similarity-based self-supervised representation learning for time series classification},
  author={Foumani, Navid Mohammadi and Tan, Chang Wei and Webb, Geoffrey I and Rezatofighi, Hamid and Salehi, Mahsa},
  journal={Data Mining and Knowledge Discovery},
  volume={38},
  number={4},
  pages={2520--2544},
  year={2024},
  publisher={Springer}
}

```