Python Machine Learning

Code samples for a directed studies course on computational modelling (UBC PSYC 547E)

This work was done primarily for learning purposes. The following have been implemented:

• Linear regression (using base Python)
• Logistic Regression (AND gate)
• Logistic Regression (OR gate)
• ANN (XOR gate)
• CNN (MNIST - hand written digits)
• CNN (CIFAR10 - colour objects)
• CNN (EEG - mind wandering classification)

Note: the code has only been tested on Windows 10 with Python 2.7.12 (64 bit)

Requirements

• Python 2.7.x
• Install: https://www.python.org/downloads/
  • This is should already be installed on Mac OSX
• numpy
  • Install: pip install numpy
  • Used to handle arrays and matrices in Python
• theano
  • Install: pip install theano
  • Used for symbolic expressions and GPU training
• matplotlib
  • Install: pip install matplotlib
  • Used for plotting
• lasagne
  • Install: pip install --upgrade https://github.com/Lasagne/Lasagne/archive/master.zip
  • This is only used for the Lasagne implementations of the different models
• nolearn
  • Install: pip install nolearn
  • This is only used for the lasagne + nolearn implementations of the different models

The quickest/easiest way to get up and running is by installing the Anaconda Python distribution, which comes with all dependencies installed (theano, lasagne, and nolearn will still need to be installed separately).