Dimensionality Reduction Techniques for Clustering Intrapersonal Violence from fMRI Data
Abstract
Intrapersonal violence (IPV) is a widespread public health issue that affects millions of people globally. Identifying patterns of brain activity resulting from IPV can aid in understanding the neural mechanisms underlying IPV and may help to develop effective interventions for preventing or coping with IPV. In this study, we explore the use of several dimensionality reduction techniques, including Principal Component Analysis (PCA), t-Distributed Stochastic Neighbor Embedding (tSNE), Uniform Manifold Approximation and Projection (UMAP), and Autoencoders, to identify clusters of brain activity associated with IPV and non-IPV subjects. We explore 3D Convolutional Neural Networks (CNNs) in combination with the Triplet Loss function to separate IPV and IPV-negative subjects, and experiment with an Explainable Artificial Intelligence algorithm to understand a classification model’s decision making. 2