The {tabnet} package, or how to outperform XGBoost and LightGBM on tabular data
Details
On Thursday, 3 February 2022 we are welcoming Christophe Regouby to RBelgium!
Christophe will present us: The {tabnet} package, or how to outperform XGBoost and LightGBM on tabular data, relying on {torch}
Google's TabNet (Arik & Pfister, 2019) is a state of the art modeling architecture for tabular data that enables interpretability and performance.
At the same time, {torch} raised Christophe's attention as being the center of a new friendly package ecosystem: the mlverse.
If you want to learn about TabNet, torch and self-supervised learning, join us online!
---
Christophe Regouby is a Data-scientist at Airbus' core AI team. He typically works on Document AI, NLP and data-visualization.
The {tabnet} package, or how to outperform XGBoost and LightGBM on tabular data