I classify a set of images using neural neural network. I am training a simple neural network with 4 inputs (image features), 1 output and 5 hidden layer with 10 nodes in it. My weight matrices are a set of numbers between -1 and 1, but I can not get a physical meaning about these weights.
Are the weights in accordance with the importance of the inputs (image features) in the model? Shouldn't I get higher weights for the inputs which are more correlated with the output? How can I get a physical meaning about the resulted weights?
Neural networks belong to so called black box models and usually it is impossible to find any interpretation of the final weights. BTW I suggest you to use a neural network with the number of outputs equal to the number of classes in your problem - each output is responsible for recognizing just one class - it is a typical approach in classification problems.
Hi, methods to understand variable contributions in NNs is explained by: Olden, J.D. and Jackson, D.A. 2002. Illuminating the ''black box'': a randomization approach for understanding variable contributions in artificial neural networks. Ecological Modelling 154 (2002) 135–150.
hello friend
yes you can use them.
Provided you have done optimization using appropriate methods.
Hope this helps you.
Thanks.
Best regards
Dr.Indrajit Mandal
- Badges
- Science topic
- Similar topics
- Computer Science
- Images