Member-only story
The summer of love for deep learning with tabular data
For many years now I have been interested in applying deep learning to tabular data. In fact, I have been so interested in this topic that I wrote a book about it. In this article, I’ll review why I think this topic is important, why I have new hope that deep learning with tabular data could go mainstream, and why the counter argument, that tabular data is fundamentally unsuitable for deep learning, is still kicking.
Why deep learning with tabular data?
Many essential aspects of modern life, from getting a loan, to using a mobile phone, to paying taxes, revolve around data that is stored in tables. Most of us have to deal with tabular data as part of our work. We live in a world of relational databases and CSV files. Deep learning has revolutionized the way we solve problems involving image data and text data. Wouldn’t it be great if the power of deep learning could have the same impact on tabular data problems?
A spate of papers on deep learning with tabular data
You can imagine my delight over the last few months as I saw a spate of papers come out on the topic of deep learning with structured data. Could 2021 be a summer of love for this topic? Recent papers that describe great results on deep learning applied to…
