We introduce UnsupervisedClustering.jl, a package that implements traditional unsupervised clustering algorithms and proposes advanced global optimization algorithms that allow escape from local optima.
In this talk, we will delve into the limitations of the traditional k-means algorithm, which often struggles to fit data that deviates from spherical distributions. In comparison, general Gaussian Mixture Models (GMMs) can fit richer structures but require estimating a quadratic number of parameters per cluster to represent the covariance matrices. Our research addresses these issues by proposing advanced global optimization algorithms that effectively combine with regularization strategies, leading to superior performance in cluster recovery compared to classical GMMs or k-means algorithms. Through a wide range of experiments on synthetic, we demonstrate the effectiveness of the proposed methods. We made available two Julia packages, UnsupervisedClustering.jl and RegularizedCovarianceMatrices.jl, that implement the proposed techniques for easy use and further research.