RCE neural network demo

Clear

Neuron Type = RBF EBF OEBF R(max) =
Neuron cnt (q) =
Output neuron cnt (m) =
Training data cnt (p) =
Epochs =

Next Step Finish vector Finish Epoch Learn Step by step STOP! Filter NN

How to use this demo

During learning process the vector, that is currently being learned is highlighted on the board as well as the neuron which it was compared to.

The board is used to draw the training data. The color represents class of data. The color can be changed by clicking on the color picker. The size of the brush can be changed by changing the number in the size input. The board can be cleared by clicking the clear button. The more gets colored the more time is needed to learn the dataset, so it is recommended to use smaller brushes.

The neural network parameters are on the right(bigger screens)/bellow(smaller screens) of the board. The first setting is the kind of Neuron used. Second is the semi-axes parameter used for initialization of neurons. Everything else represents the state of the network.

The buttons bellow the board control the learning process

• Next Step - moves the learning process one step forward (either moves to next vector or tests the next neuron)
• Finish vector - finishes learning the current vector (one vector from the dataset applied to every neuron in hidden layer)
• Finish Epoch - finishes learning the current epoch (the whole dataset one time)
• Learn - finishes all epochs needed to fully learn the training data
• Step by Step - shows the learning process in human time
• Stop - stops the learning process
• Filter NN - use after fully learning the training data, filters out neurons that are obsolete