Topological structure of complex predictions

TL;DR

This paper introduces a topological data analysis approach to visualize and interpret complex prediction models like deep learning, aiding in understanding, error detection, and insights across various scientific domains.

Contribution

It applies topological data analysis to complex models, providing scalable visualization tools for interpretation and error detection in large datasets.

Findings

Enabled detection of labeling errors in training data

Provided insights into model generalization in image classification

Inspected predictions of pathogenic mutations in BRCA1 gene

Abstract

Complex prediction models such as deep learning are the output from fitting machine learning, neural networks, or AI models to a set of training data. These are now standard tools in science. A key challenge with the current generation of models is that they are highly parameterized, which makes describing and interpreting the prediction strategies difficult. We use topological data analysis to transform these complex prediction models into pictures representing a topological view. The result is a map of the predictions that enables inspection. The methods scale up to large datasets across different domains and enable us to detect labeling errors in training data, understand generalization in image classification, and inspect predictions of likely pathogenic mutations in the BRCA1 gene.