Mutilayer perceptron (feedforward neural) network with a single layer of
hidden units, trained using weight decay as a regularizer on the weights. The
motivation is to restrict overfitting by keeping the weights small. In
conventional function fitting this method is called Ridge Regression and is a
fairly standard text book forms of regularization [1]. The details of the
procedure used to generate the results are reported
here
Software
The source is available as a compressed
tar file. However, be forewarned that this source is not stand-alone. It
runs within the neural net simulator Uts 4.1 developed at the University
of Toronto. The simulator is available free of charge. You
will first have to install it. The simulator will only run on UNIX like
systems and requires Tcl, see the website for details. Only the Tcl
scripts used for running the simulator are include in the tar file.
Results
Directory listing of the results available for the
mlp-wd-1 method. Put the desired files in the appropriate methods
directory in your delve hierarchy and uncompress them with using the
"gunzip *.gz" command and untar them using "tar -xvf
*.tar".
Related References
[1] C. M. Bishop. Neural networks for pattern recognition Oxford
University Press 1995.