iCubed (Institute, Industry, Innovation) seminars invite DSI Industry Affiliates to give technical talks on work going on in their domain. Join to learn about real-world uses of data science and opportunities with Industry Affiliates.

Guest Speaker

Bayan Bruss, Sr. Director, Applied ML Research at Capital One

Moderated By: John Hyde, Assistant Director of Career Development and Alumni Services, DSI

Details

Monday, December 13 (2:00 PM – 3:00 PM ET) – Virtual

Talk Information

Deep Learning for Tabular Data

Tabular data is the dominant paradigm for machine learning models in a number of industries such as financial services. Despite this importance, we have yet to see the performance that deep neural networks have shown in other domains like language and vision. Methods like Gradient Boosted Machines and Random Forests still provide a very strong baseline for most tabular datasets. To-date this baseline has not been convincingly exceeded by neural methods. In this presentation Bayan will be giving a brief introduction to Applied Research at Capital One and highlighting research that we have been doing with researchers at University of Maryland. We have developed a method called SAINT that uses Transformer architectures for tabular data and surpass GBMs and Random Forests on a broad number of datasets.