Download the .tgz bundle of MATLAB code

Requirements

  1. MATLAB
  2. MATLAB Lightspeed toolbox

Scripts

BICfn.m Calculating the Bayesian Information Criteria.
Bin_Class_EM_fn4.m Implements the EM algorithm for Bayesian classification with averaged feature clusters.
Bin_Class_EM_loop.m Runs Bin_Class_EM_fn4.m multiple times and stores the results.
Bin_Exps_Loop.sh An example shell script for submission to a grid system.
GammaUpdate3.m Update to γ in the E-step of the EM algorithm.
LogJoint3.m Calculating the log joint likelihood.
PiUpdate.m Update to π in the M-step of the EM algorithm.
SigmaSqdUpdate3.m Update to σ in the M-step of the EM algorithm.
ThetaUpdate.m Update to θ in the M-step of the EM algorithm.
wUpdate.m Update to w in the M-step of the EM algorithm.
yUpdate.m Update to y in the E-step of the EM algorithm.

Instructions

  1. Prepare your data appropriately (as described in [1]);
  2. Download .tgz bundle;
  3. Set the values of lightspeed_dir, working_dir, data_file, dataset_string in Bin_Class_EM_loop.m to correspond to your file structure.
  4. Set the value of data_file in Bin_Class_EM_fn4.m to match the value in Bin_Class_EM_loop.m.
  5. If distributing across a cluster, use Bin_Exps_Loop.sh as a template, making sure you refer to the correct version of MATLAB (i.e., give the full path to the executable).