Skip to content

Latest commit

 

History

History
68 lines (39 loc) · 1.85 KB

nn.md

File metadata and controls

68 lines (39 loc) · 1.85 KB
Raw

Neural Network

The Neural Network is a simple feedforward neural network implementation in JavaScript. It consists of methods for creating, training, testing, saving, and loading a neural network model.

Class: NeuralNetwork

Represents a feedforward neural network with customizable architecture and learning parameters.

Constructor

Creates a new instance of the NeuralNetwork class.

new NeuralNetwork(inputSize, hiddenSize, outputSize, learningRate)
  • inputSize (number): The number of input neurons.
  • hiddenSize (number): The number of neurons in the hidden layer.
  • outputSize (number): The number of output neurons.
  • learningRate (number): The learning rate for weight updates during training.

Methods

.sigmoid(x)

Calculates the sigmoid activation function value for a given input.

  • x (number): The input value.
  • Returns: The sigmoid output.

.feedforward(inputs)

Performs a feedforward pass through the neural network.

  • inputs (number[]): The input values.
  • Returns: The output values.

.backpropagation(inputs, targets)

Performs backpropagation to update the network's weights based on error.

  • inputs (number[]): The input values.
  • targets (number[]): The target output values.

.train(inputs, targets)

Trains the neural network using the provided training data.

  • trainingData (Array<[number[], number[]]>): An array of input-target pairs for training.
  • numberOfIterations (number): The number of training iterations.

.test(input)

Tests the neural network using the provided input and displays the output.

  • input (number[]): The input values.

.saveModel(filePath)

Saves the model's weights to a JSON file.

  • filePath (string): The file path to save the model to.

.loadModel(filePath)

Loads the model's weights from a JSON file.

  • filePath (string): The file path to load the model from.