SONG: Self-Organizing Neural Graphs

TL;DR

SONG introduces a novel gradient-based training method for decision graphs, enabling deep interpretable neural networks that outperform or match existing models on various datasets.

Contribution

The paper presents Self-Organizing Neural Graphs (SONG), a new decision graph model trained efficiently via a Markov process-based paradigm, advancing deep interpretable neural network design.

Findings

SONG performs on par or better than existing decision models.

Theoretical analysis supports the effectiveness of SONG.

Experimental results on multiple datasets validate the approach.

Abstract

Recent years have seen a surge in research on deep interpretable neural networks with decision trees as one of the most commonly incorporated tools. There are at least three advantages of using decision trees over logistic regression classification models: they are easy to interpret since they are based on binary decisions, they can make decisions faster, and they provide a hierarchy of classes. However, one of the well-known drawbacks of decision trees, as compared to decision graphs, is that decision trees cannot reuse the decision nodes. Nevertheless, decision graphs were not commonly used in deep learning due to the lack of efficient gradient-based training techniques. In this paper, we fill this gap and provide a general paradigm based on Markov processes, which allows for efficient training of the special type of decision graphs, which we call Self-Organizing Neural Graphs (SONG). We provide an extensive theoretical study of SONG, complemented by experiments conducted on Letter, Connect4, MNIST, CIFAR, and TinyImageNet datasets, showing that our method performs on par or better than existing decision models.