search

jacobgil / pytorch-grad-cam

Advanced AI Explainability for computer vision. Support for CNNs, Vision Transformers, Classification, Object detection, Segmentation, Image similarity and more.
https://jacobgil.github.io/pytorch-gradcam-book
MIT License
10.4k stars 1.55k forks source link

Can it be used for one-dimensional data? For example, time series data #518

Open adrenaline0425 opened 3 months ago

adrenaline0425 commented 3 months ago

Can it be used for one-dimensional data? For example, time series data

iganggang commented 3 months ago

@adrenaline0425 @Kaushalya @mingloo @karray @rachellea have the same question. i wang to visualize attention map in time series. if u have any news, please kick me. Thank you very much.

karray commented 3 months ago

@iganggang @adrenaline0425 this library primarily focuses on CAM methods, which are specifically designed for convolutional networks.

Consider exploring model agnostic methods (e.g., RISE, SHAP, LIME), gradient methods (Integrated Gradients), or attribution methods such as LRP.

iganggang commented 3 months ago

@karray thank you very much. i can not plot attention map without examples according to my poor level. i give up it.

karray commented 3 months ago

@iganggang take a look at AttnLRP

iganggang commented 3 months ago

@karray thanks a lot. your abilities are very strong. hope you live a peaceful life.