GPML/readme.txt at master · rexsutton/GPML · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Rex Sutton 2016

This is the `readme' for my python implementation of binary classification, using a Gaussian Process with Laplace's approximation.
Included is a demonstration taken from `Rasmussen, Williams 2006 Gaussian Processes For Machine Learning', see `http://www.gaussianprocess.org/gpml/'.
The demo uses machine learning to classify images of hand-written digits collected by United States Postal Service, in particular images of `3' and `5's.
The USPS data is taken from `http://www.gaussianprocess.org/gpml/data/'.

setup.sh
--------

Run this script first to download and de-compress the demo data.

demo.sh
--------

Run this script in the shell to view the demo.
The model is trained and then tested interactively.
Training should take less than two minutes to complete on a modern laptop.