SPIRAL

Environment setup with uv

uv python install 3.10.13
uv python list
uv init --python 3.10.1
uv venv --python 3.10.13
source .venv/bin/activate
python -V
python -m ensurepip --upgrade
python -m pip install --upgrade pip
pip -V
uv sync
uv tool install ruff

Data

Download CSV files from https://www.thedatum.org/datasets/TSB-AD-U.zip and place them in data/TSB-AD/TSB-AD-U/.

All datasets used in this study are publicly available.

Repository structure

1. cli/ — CLI commands for data transformation pipelines (time series to images) and vision backbone training
2. configs/ — configuration parameters and hyperparameters used in this study
3. critical_difference/ — critical difference plots with per-metric rankings between time-domain baselines and vision models trained on TS2I representations (generated by critical_difference.py)
4. plgrid/ — execution scripts for HELIOS supercomputer and PLGrid infrastructure (SLURM)
5. results/ — TS2I training and evaluation results with processing, aggregation and visualization scripts
6. scripts/ — helper scripts
7. src/ — source code
8. ts2i_metadata/ — metadata collected during time series to image processing
9. ts2i_visualization/ — sample anomaly visualizations using all 10 TS2I techniques (including SPIRAL)
10. TSB-AD-results/ — time-domain baseline evaluation results with processing, aggregation and visualization scripts
11. TSB_AD/ — module sourced from the original TSB-AD benchmark repository to ensure consistent evaluation protocol

Citation

This research work is currently under submission to a peer-reviewed journal.

License

⚠️ This code and methodology are part of a research project currently under peer review. Do not copy or redistribute without proper citation or prior permission from the authors.

Contact

For inquiries, contact smendowski@agh.edu.pl.

Author: Mateusz Smendowski