MlpModel is carrying structure of Multilayer Perceptron Classifier, which consists of input, hidden and output layers. Visualization of MlpModel' structure can be seen below:

MlpModel can exists in three states:

1. Uninitialized - MlpModel is in this state before call to MLP_Init.
2. Not trained - MlpModel is initialized, but not trained. Next step to achieving a useful MlpModel is to pass it to MLP_Train.
3. Trained - MlpModel is ready to perform classification. It can be passed to MLP_Respond.

MlpModel can only be used with Multilayer Perceptron category filters. Who wants to use MLP to perform optical character recognition operation should use OcrMlpModel and suitable Optical Character Recognition filters.