Matrix Manifold Optimization for Gaussian Mixtures

Authors: Reshad Hosseini, Suvrit Sra

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We have performed numerous experiments to examine effectiveness of our method. Below we report performance comparisons on both real and simulated data. In all experiments, we initialize the mixture parameters for all methods using k-means++ [2].
Researcher Affiliation Academia Reshad Hosseini School of ECE College of Engineering University of Tehran, Tehran, Iran reshad.hosseini@ut.ac.ir Suvrit Sra Laboratory for Information and Decision Systems Massachusetts Institute of Technology Cambridge, MA. suvrit@mit.edu
Pseudocode Yes Algorithm 1: Sketch of an optimization algorithm (CG, LBFGS) to minimize f(X) on a manifold
Open Source Code Yes To aid reproducibility of our results, MATLAB implementations of our methods are available as a part of the MIXEST toolbox developed by our group [12].
Open Datasets Yes Available at UCI machine learning dataset repository via https://archive.ics.uci.edu/ml/datasets
Dataset Splits No The paper does not explicitly provide details about training, validation, or test dataset splits.
Hardware Specification No The paper does not provide any specific hardware details used for running its experiments.
Software Dependencies No The paper mentions 'MATLAB implementations', 'MIXEST toolbox' [12], and 'MANOPT' [6], but does not provide specific version numbers for these software components.
Experiment Setup Yes In all experiments, we initialize the mixture parameters for all methods using k-means++ [2]. All methods also use the same termination criteria: they stop either when the difference of average log-likelihood (i.e., 1/n log-likelihood) between consecutive iterations falls below 10^-6, or when the number of iterations exceeds 1500.