perceptron

README

Perceptron

A linear classifier written in Go to demonstrate the fundamentals of supervised learning.

Documentation

Overview

Package perceptron provides the ability to create, train and use a linear classifier which takes a number of inputs and returns a single output.

A Perceptron has a set of input weights and an activation threshold.

For a given set of inputs i1...in, the activity is calculated as

sum(i * w)

where w1...wn are weights for the inputs. The output is 1 if the activity > threshold, else 0.

Index

• type Perceptron
  • func New(numInputs int) *Perceptron
  • func (p *Perceptron) Activates(input []int) int
  • func (p *Perceptron) Activity(input []int) float64
  • func (p *Perceptron) Train(input []int, expectedOutput int)

Types

type Perceptron

type Perceptron struct {
	Weights   []float64
	Threshold float64
}

func New

func New(numInputs int) *Perceptron

New creates a new Perceptron with the specified number of inputs. It assigns random weights to each of the inputs and generates a random activation threshold for the Perceptron.

func (*Perceptron) Activates

func (p *Perceptron) Activates(input []int) int

Activates checks if the Perceptron activates for the given input. It returns 0 or 1 depending on the input and how the perceptron has been trained.

func (*Perceptron) Activity

func (p *Perceptron) Activity(input []int) float64

Activity calculates the activity on the Perceptron for a given input. It doesn't calculate if the Perceptron actually fires, see "Activates" instead.

func (*Perceptron) Train

func (p *Perceptron) Train(input []int, expectedOutput int)

Train takes a given input and an expected output. If the Perceptron doesn't activate, it adjusts the activation threshold and input weights based on the delta between the expected and calculated output.