CODEWITHSUNDEEP
javaneural-network
user974227
user974227

Reputation: 209

Testing Multilayer Perceptron

I'm very new to Neural Networks and I'm still trying to understand the concepts. I've seen implementations of MLP in different programming languages. According to what I've read, the goal is to find the weights which will make the outputs match the expected.

What if I already have the weights from the trained network? How do I make use of those weights to test the network? Say I have new inputs for the testing set, how do I use the trained network to look for matches from the training set? (I'm doing image recognition) There are available libraries but I want to at least know how things work. I've read and scanned through a few but I can't seem the find where the testing method is.

I'm reading articles about MLP but I still lack knowledge. I badly want to understand this part. Thanks in advance.

Upvotes: 0

Views: 1304

Answers (1)

MrSmith42
MrSmith42

Reputation: 10151

The idea is supervised learning with the training set. The Network learns (the weights) (e.g. by backpropagation) to classify the training set correctly.

When you present a new input to your network, the hope is that the learned weights lead to a correct classification of the new input.

Upvotes: 1

Related Questions