Neuron

For this project, I learned a bunch of python and used it, as well as Machine Learning: An Algorthmic Perspective by Stephen Marsland, to program a neuron that calculates the correct output of logic functions, such as AND, OR, and XOR, based on a set of input data. The neuron is fed a data set of values for two points, such as (0,0), (0,1), (1,0), (1,1), and learns the correct output by using formulas to calculate a proposed output based on some randomized weights. The weights are updated after every iteration to more accurately reflect the output. Also, graphs are generated of the error in the proposed outputs generated by the neuron over each iteration. Sometimes, the neuron does not find the completely correct weights in the given limit of iterations (this is expected). Some functions that are more difficult to calculate than others (such as XOR, which requires three dimensional data) are allowed more time to iterate. I am very interested in artificial intelligence, although I am not experienced in it at all, so I wanted to do something AI-related for my hack. It was extremely interesting, and I am glad I had this opportunity to learn it.

Built With

• eclipse
• numpy
• pydev
• pylab
• python27
• windows