SuperTML: Domain Transfer from Computer Vision to Structured Tabular Data through Two-Dimensional Word Embedding

Anonymous

Keywords: Structured Tabular Data, CNN Model, Transfer Learning, Two-dimensional Word Embedding, Pre-trained ImageNet Model

Abstract: Structured tabular data is the most commonly used form of data in industry according to a Kaggle ML and DS Survey. Gradient Boosting Trees, Support Vector Machine, Random Forest, and Logistic Regression are typically used for classification tasks on tabular data. The recent work of Super Characters method using two-dimensional word embeddings achieved state-of-the-art results in text classification tasks, showcasing the promise of this new approach. In this paper, we propose the SuperTML method, which borrows the idea of Super Characters method and two-dimensional embeddings to address the problem of classification on tabular data. For each input of tabular data, the features are first projected into two-dimensional embeddings like an image, and then this image is fed into fine-tuned ImageNet CNN models for classification. Experimental results have shown that the proposed SuperTML method have achieved state-of-the-art results on both large and small datasets.

TL;DR: Deep learning on structured tabular data using two-dimensional word embedding with fine-tuned ImageNet pre-trained CNN model.