Neural Networks: Minimal, open source example with exhaustive training data?

Question

I am currently searching for a neural network (toy) example, that I might optimize using GPU kernels. I need

• clean & minimal code (no fancy optimizations, max. a few thousand sloc)
• ample testing data
• C/C++ open source

Does anybody has an adequate hint?

There's a bunch of examples on the net, but with insufficient testing data (there's no sense in GPU-optimizing code, that executes in a few seconds anyway)

Answer 1

clean & minimal code (no fancy optimizations, max. a few thousand sloc) C/C++ open source

The venerable Neural Networks at your Fingertips (or archived here ) is written very cleanly in C, it's very easy to play with and contains implementations for a number of 'popular' architectures (since you don't specify what neural network architecture(s) you're interested in).

ample testing data

You can find ample testing data (amongst other places) at the UC Irvine Machine Learning Repository.