Neurosymbolic models based on hybrids of convolutional neural networks and decision trees

Rasul Kairgeldin, Miguel Á. Carreira-Perpiñán

Keywords: convolutional neural networks, decision trees, hybrid models, interpretable AI

Track: Main Track

Abstract: Building on previous work, we propose a specific form of neurosymbolic model consisting of the composition of convolutional neural network layers with a sparse oblique classification tree (having hyperplane splits using few features). This can be seen as a neural feature extraction that finds a more suitable representation of the input space followed by a form of rule-based reasoning to arrive at a decision that can be explained. We show how to control the sparsity across the different decision nodes of the tree and its effect on the explanations produced. We demonstrate this on image classification tasks and show, among other things, that relatively small subsets of neurons are entirely responsible for the classification into specific classes, and that the neurons' receptive fields focus on areas of the image that provide best discrimination.

Paper Type: Long Paper

Submission Number: 49