Lightweight Neural Network++ documentation

Visit the lwnnplus home page, download the technical report in english or italian.

Lightweight Neural Network ++ Documentation

0.996

About the documentation

This is the documentation for the developer, and includes all the interfaces of the classes. If you want a technical report and a collection of examples, download the report (ps.gz) (also in italian)

How to start to use LightWeight Neural Network++ for your project

Let lwnn++ understand your data

If you just want to train a network to do some work you can use the standard tools provided by the project, like lwnn_visualtrainer. You have to provide the data in lwnn or fann format for training (see iomanagelwnn for the description of the format of these files) If it's not easy to convert the data in this format you can write your own class derived of iomanage that reads in the proper format the number of patterns, of input and output (and the whole data!) from your own files. You will find some examples of how to write this derived class in the directory examples/

Then you can use method convert of class iomanagebinary to save your data in binary format and use it from standard lwnn_visualtrainer.

Train your network

You can use the standard lwnn_visualtrainer to train the network as you need it for your data. When the net is properly trained you can save it on a file and use it from your own program

Use the network

To use the network you have just to include <network.h> (and maybe you cand just link with network.o and not with the whole library), then load your network from a file and use the method network::compute(input, output) to get your results. That's all!

---