We are introducing a saliency map method that produces meaningful attributions for time-series ML models. Instead of restricting explanations to the raw time domain, our approach generates saliency maps in relevant domains such as frequency or ICA components - improving interpretability when applying deep model to neural time-domain signals such as EEG.
Why it matters
Standard attributions methods (e.g. Integrated Gradients) yield point-wise importance in the time domain, which can be difficult to interpret in neuroscience. Our method extends this idea to domains that are more aligned with scientific practice.
Key Features
Cross-Domain Saliency Maps is an open-source toolkit that:
- Provides frequency and ICA domain attributions out-of-the-box
- Extends to an invertible transform with a differentiable inverse
- Works plug-and-play with PyTorch and TensorFlow, no retraining required
- Demonstrates utility on EEG seizure detection and other datasets
Example outputs
Cross-domain attributions reveal more interpretable patterns by surfacing features in frequency bands and in independent components. For more details on the plot check our Colab demo!
Try it out:
- Whatdoes your deep model see in your EEG? (Colab demo using MNE) Google Colab
- Get the code & run it on your models (GitHub repo) GitHub - esl-epfl/cross-domain-saliency-maps: Pytorch/Tensorflow package for generating saliency maps for time-series models using Cross-Domain Integrated Gradients.
- Read the full story (arxiv preprint) https://arxiv.org/pdf/2505.13100
We’d be happy to hear your thoughts and experiences applying this to time-series neural data.