plsomlib

Open source implementation of the PLSOM and PLSOM2 algorithms. Implementations in Java, C#, and C++.

Project organisation.

The code is split into three subdirectories:
cpp/
csharp/
java/

Each contains a separate language implementation.

The Java implementation is the most complete one. It also contains a lot of vaguely related code that is only of interest to Machine Learning researchers.

The C# implementation contains a basic implementation of the PLSOM2 algorithm, nothing else. There is no code for saving and loading a map.

The C++ implementation is in progress.

Usage

Regardless of language, the use of a PLSOM2 is similar.

Java example:

// Create an object.
// This example has 60-dimensional input and three output dimensions in a 20x20x20 cube.
PLSOM2 net = new PLSOM2(60,20,20,20);
// set the generalisation factor
net.setNeighbourhoodRange (30);
// Train the network.
for(double [] data:trainingData) 
{
    net.train(data);
}
// Use the network to classify input.
int [] classification  = net.classify(input);

C++ example:

// Create an object.
// This example has 60-dimensional input and three output dimensions in a 20x20x20 cube.
PLSOM2 net(60,{20,20,20});
// Set the generalisation factor.
net.setNeighbourhoodRange(30);
// Train the network.
for(int x = 0;x<data.size();x++)
{
  net.train(data[x]);
}
// Use the network to classify input.
vector<int> classification = net.classify(input);