UCSL : A Machine Learning Expectation-Maximization framework for Unsupervised Clustering driven by Supervised Learning

TL;DR

UCSL is a versatile EM-based framework that combines supervised learning with clustering to discover meaningful subgroups, validated on synthetic and real datasets with improved accuracy over existing methods.

Contribution

The paper introduces UCSL, a novel EM ensemble method that integrates supervised models into clustering, capable of handling classification and regression tasks with a new dimension-reduction technique.

Findings

Achieves +1.9 points in balanced accuracy over state-of-the-art.

Effectively identifies consistent subtypes in psychiatric disease data.

Demonstrates robustness and generalization on synthetic and real datasets.

Abstract

Subtype Discovery consists in finding interpretable and consistent sub-parts of a dataset, which are also relevant to a certain supervised task. From a mathematical point of view, this can be defined as a clustering task driven by supervised learning in order to uncover subgroups in line with the supervised prediction. In this paper, we propose a general Expectation-Maximization ensemble framework entitled UCSL (Unsupervised Clustering driven by Supervised Learning). Our method is generic, it can integrate any clustering method and can be driven by both binary classification and regression. We propose to construct a non-linear model by merging multiple linear estimators, one per cluster. Each hyperplane is estimated so that it correctly discriminates - or predict - only one cluster. We use SVC or Logistic Regression for classification and SVR for regression. Furthermore, to perform cluster analysis within a more suitable space, we also propose a dimension-reduction algorithm that projects the data onto an orthonormal space relevant to the supervised task. We analyze the robustness and generalization capability of our algorithm using synthetic and experimental datasets. In particular, we validate its ability to identify suitable consistent sub-types by conducting a psychiatric-diseases cluster analysis with known ground-truth labels. The gain of the proposed method over previous state-of-the-art techniques is about +1.9 points in terms of balanced accuracy. Finally, we make codes and examples available in a scikit-learn-compatible Python package at https://github.com/neurospin-projects/2021_rlouiset_ucsl