Class

Neural

Neural(fname)

Constructor

# new Neural(fname)

Create a artificial neural network.

Note: Neural module must be loaded by calling LoadLibrary("neural") before using!

Parameters:
Name Type Description
fname string

file name of training data previously stored by Save()

See:
  • LoadLibrary()
  • Save()

Methods

# Close()

free ressources of the neural network.

# Run(inp) → {Array.<number>}

Run the network on some input and get a response.

Parameters:
Name Type Description
inp Array.<number> | DoubleArray

Input data. must contain at least the number of input values as specified when creating the ANN.

The networks response to the input.

Array.<number>

# Save(fname)

Store training data of the network to disk.

Parameters:
Name Type Description
fname string

The file name.

# Train(inp, outp, rate)

Train the network with a single dataset.

Parameters:
Name Type Description
inp Array.<number> | DoubleArray

Input data. Must contain at least the number of input values as specified when creating the ANN.

outp Array.<number> | DoubleArray

Expected output data. Must contain at least the number of output values as specified when creating the ANN.

rate number

Learning rate (>0).

Neural(inputs, num_hidden, hidden, outputs)

Constructor

# new Neural(inputs, num_hidden, hidden, outputs)

Create a artificial neural network.

Note: Neural module must be loaded by calling LoadLibrary("neural") before using!

Parameters:
Name Type Description
inputs number

size of input layer

num_hidden number

number of hidden layers

hidden number

size of the hidden layers

outputs number

size of output layer

See:
  • LoadLibrary()

Methods

# Close()

free ressources of the neural network.

# Run(inp) → {Array.<number>}

Run the network on some input and get a response.

Parameters:
Name Type Description
inp Array.<number> | DoubleArray

Input data. must contain at least the number of input values as specified when creating the ANN.

The networks response to the input.

Array.<number>

# Save(fname)

Store training data of the network to disk.

Parameters:
Name Type Description
fname string

The file name.

# Train(inp, outp, rate)

Train the network with a single dataset.

Parameters:
Name Type Description
inp Array.<number> | DoubleArray

Input data. Must contain at least the number of input values as specified when creating the ANN.

outp Array.<number> | DoubleArray

Expected output data. Must contain at least the number of output values as specified when creating the ANN.

rate number

Learning rate (>0).