Objectives:

Effectively communicate your findings
(Extra Credit) Design and test your own code to extend our project
Specification:

You should submit a Word or pdf document with a discussion of the results of each run. Feel free to alter the topology and learning rate of the networks, or the number of epochs used to train. Comment on how these changes affect the results. Include a graph of the sin curve produced by the test method superimposed on an actual sin graph from 0 to pi/2. You can use any tool you like to produce the graph. Discuss where the network succeeded and failed, and come up with some possible reasons for the failure.

You may also complete one or more of the following extra credit opportunities. You must produce high-quality work with reasonable commentary to receive extra credit. It is not sufficient to paste the output of your code.
- - - -
Train the XOR function with and without a bias input.

For each scenario:

Train five times to 20,000 epochs each time, recording RMSE at each 100 epochs. Average the results (i.e. average the five values at epoch 100, epoch 200, etc)

In your favorite application or package, graph the training progress of each scenario together on one graph (epochs on the x-axis, RMSE on the y-axis). Discuss what you see, and why.



PLEASE HELP ITS DUE ON MARCH 27!