Formally we want to construct a neural network to

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: irst one evaluting the output of the network given the input, and a second one, implementing the weight update. Let’s first define a network structure for the problem above: ± Ò Ò ÒÔÙØ Ò Ò ÒÓÙØÔÙØ Ò ± Ò Ò Ò Ø ¾¼ ½ Ò ØÛÓÖ ÓÖ Ø Ö Ð ×× Ø ÓÒ ÔÖÓ Ð Ñ ØÛÓÖ ß Ö Ø ØÖ Ð Ý Ö ¼ ÒÔÙظ ½ ØÛÓÖ ºÐ Ý Ö¼ Þ ÖÓ×´Ò ÒÔÙظ½µ ØÛÓÖ ºÐ Ý Ö½ Þ ÖÓ×´Ò Ò¸½µ ØÛÓÖ ºÐ Ý Ö¾ Þ ÖÓ×´ÒÓÙØÔÙظ½µ Ò¸ ¾ ÓÙØÔÙØ ± ÈÙØ Ö Ò ÓÑ ×Ñ ÐÐ Ú ÐÙ × ÓÖ Ø Û Ø× Ò ØÛÓÖ ºÛ ؽ Ö Ò Ò´Ò Ò¸Ò ÒÔÙØ·½µ Ò ØÛÓÖ ºÛ ؾ Ö Ò Ò´ÒÓÙØÔÙØ¸Ò Ò·½µ Based on this network definition, we can write a function to evaluate the network output: ± Ò ± Ò ÙÒ Ø ÓÒ Ò Ø Ú ÐÙ Ø ÒÒ´ ÒÔÙØ׸ Ò ØÛÓÖ µ Ì Ö×Ø Ð Ý Ö × × Ø ØÓ Ø ÒÔÙظ ÔÐÙ× ØÛÓÖ ºÐ Ý Ö¼ ½ ÒÔÙØ×℄ Ì Ò Ð Ý Ö × ÓÑÔÙØ Ù× Ò Ø ØÒ Ø½ Ò ØÛÓÖ ºÛ ؽ ¶ Ò ØÛÓÖ ºÐ Ý Ö¼ ÓÒ×Ø ÒØ ÙÒ Ø Ø Ú Ø ÓÒ 6...
View Full Document

This note was uploaded on 03/21/2010 for the course IT ANN taught by Professor Mario during the Spring '10 term at École Normale Supérieure.

Ask a homework question - tutors are online