Our method-related code is in the folder of models and pre-processing related code is in the folder of data_provider.
To run our code please install PyTorch with cuda support. For our package, we have used 1.13.0 version. It can be installed from this link. We used CUDA 11.7 option.
Please also install Mamba. The installation procedures are mentioned here. Please consider using a Linux system such as Ubuntu for smoother compilation. It might face some issues while installing on other systems (e.g., issue1, issue2, etc.) In addition to those mentioned above, we also used several other packages mentioned in requirements.txt files.
Download the datasets from the official website in .ts format. Place the downloaded files in the datasets/X folder, where X is the dataset name (e.g., SpokenArabicDigits, Handwriting, etc.).
To run for a dataset please use this command sh ./scripts/classification/TSCMamba.sh, for example this will run for SpokenArabicDigits dataset and will generate relevant checkpoint, result, etc. Modify it as per your requirements.
We are deeply grateful for the valuable code and efforts contributed by the following GitHub repositories. Their contributions have been immensely beneficial to our work.
- Mamba (https://github.com/state-spaces/mamba)
- Time-Series-Library (https://github.com/thuml/Time-Series-Library)
We also thank the data creators and curators for their hard work in making these datasets publicly available.
If you find our work useful in your research, please consider citing our paper as follows:
@article{tscmamba,
title = {TSCMamba: Mamba meets multi-view learning for time series classification},
journal = {Information Fusion},
volume = {120},
pages = {103079},
year = {2025},
issn = {1566-2535},
doi = {https://doi.org/10.1016/j.inffus.2025.103079},
url = {https://www.sciencedirect.com/science/article/pii/S1566253525001526},
author = {Md Atik Ahamed and Qiang Cheng},
}