Skip to content

[ICML 2024 (Spotlight)] InterpreTabNet: Distilling Predictive Signals from Tabular Data by Salient Feature Interpretation. Paper: https://arxiv.org/abs/2406.00426.

License

Notifications You must be signed in to change notification settings

jacobyhsi/InterpreTabNet

Repository files navigation

InterpreTabNet: Distilling Predictive Signals from Tabular Data by Salient Feature Interpretation

MIT License Paper URL OpenReview

Model Logo

Figure 1: The InterpreTabNet Architecture presenting a variational formulation of the TabNet encoder.

Model Logo

Figure 2: Left (a): Learned masks associated with InterpreTabNet. Right (b): Learned masks associated with TabNet. Bottom (c): Stacked InterpreTabNet Feature Masks between subsequent feature masks.

Usage

Clone this repository and navigate to it in your terminal. Install required packages and dependencies as follows.

conda create -n interpretabnet python=3.10
conda activate interpretabnet
pip install -r requirements.txt

To run InterpreTabNet with a desired dataset, it is recommended to use the "interpretabnet.ipynb" file for the most up-to-date codebase.

About

[ICML 2024 (Spotlight)] InterpreTabNet: Distilling Predictive Signals from Tabular Data by Salient Feature Interpretation. Paper: https://arxiv.org/abs/2406.00426.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published