Using this two function the network

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: performance is very sensitive to the initial random conditions. In my observations, in this specific example, around 5% performance variation can be expected when varying the random seed. 5 Are your conclusions to the previous question sensitive to the initial random conditions? The selection of the least and most important characteristics is very sensitive here to the random seed. If a features was really useless, then it uselessness would not depend on the random seed. This is a hint that all theses features are useful to classify crabs. Problem 3: Implementation of the backpropagation algorithm In the class, we have seen how to implement the back-propagation algorithm using the vector based representation of the neural network. Using the crab classification data set, the algorithm described in the slides, and a network with 20 internal nodes, implement the back propagation algorithm. How does your implementation compares to Matlab toolbox implementation in terms of classification performances? In order to implement neural network training, we need to implement two functions: a f...
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