University of Birmingham > Talks@bham > Astrophysics Talks Series > Artificial Neural Networks and Data Analysis

Artificial Neural Networks and Data Analysis

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Matteo Bianconi.

I will speak about pattern recognition using Artificial Neural Networks (ANN) and SOM (SOM) ( self .organizing maps). First I define the input and output patterns and the search of the dimension of the patterns. This requires to look for the correlation length of the time series.. I will define the training, validation and learning error. The dimension of inputs patterns is defined by the correlation length. The architecture is defined also by the number of input neurons, layers and synaptic weights. The best architecture is found minimizing the learning and generalization error studying the variation of the dependence of these quantities on the possible choices a heuristic research. I will show examples of networks with exact estimates of the learning error. The minimization of the error can be performed ,both theoretically and experimentally only if the number of patterns is large. The case in which is possible to show rigourosly the theoretical result holds only for an infinite number of patterns and infinite number of neurons.

This talk is part of the Astrophysics Talks Series series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


Talks@bham, University of Birmingham. Contact Us | Help and Documentation | Privacy and Publicity.
talks@bham is based on from the University of Cambridge.