DM825 - Introduction to Machine Learning
Sheet 5, Spring 2013 [pdf format]



Exercise 1

Suppose you had a neural network with linear action functions. That is, for each unit the output is some constant c times the weighted sum of the inputs.




Exercise 2

Using the data from the previous exercises, investigate the use of nnet from the homonymous package in R to learn a neural network. The function provides an implementation to fit single-hidden-layer neural networks, possibly with skip-layer connections (i.e., a link from the input node directly to the output nodes). In particular, use that function for both regression and binary classification and become acquainted with the changes that they the two cases require. For multinomial classification try the function multinom from the same function. Start by looking at the examples. Finally, compare the performance against the previous linear regression models via glm and mlogit.