Applying Deep Learning on Tabular data for Regression and Classification problems

Applying DL to structured data via categorical embeddings with FastAi

It’s a common sentiment that Deep Learning is only good for images and language models. This post is about using Deep Learning on tabular data, for both Regression and Classification problems. We will use FastAi library for creating our deep learning models. We will use Kaggle competitions as benchmarks to see how our solutions compares to other solutions using traditional ML models.

If you haven’t watched FastAi tutorials already, please visit this link for the awesome and free tutorials.

Network architecture

Here is a quick view of the network I have in mind. We will use FastAi to transform this vision to a real network.

Steps/Layers:

1. Categorical embeddings: Similar to latent features, embedding categories into N-dimensional features.
2. Continuous variables: Batch Normalisation for continuous variables
3. Hidden layers