Loss functions for fast dropout/dropout and baselines like softmax, and logistic regressions They are described in the readme file in the respective folders. To get the idea behind them, see the fast dropout paper on my website at stanford.edu/~sidaw
If you have minFunc, you can try running run_experiment.m The example_data loaded are vectors from the atheism vs. christian newsgroup task. With the fixed random seed, I got: DetDropout: 0.882188 Dropout: 0.882188 LR: 0.827489
Dependency on minFunc:
minFunc can be found here: http://www.di.ens.fr/~mschmidt/Software/minFunc.html