Perceptron

• Perceptron (html & JavaScript)
• Perceptron.html

Introduction

• Single perceptron
• No activation function
• Two inputs, two train data, bias input (optional), learning rate
• Hyperplane is shown

Usage

• Start: generates two train data (p1 and p2, desired output (d): 1 and -1) with two featrues ranging [0, 1] , and resets weights
• Step:
  • trains by each train data, alternately
  • updates by w₁(t+1) = w₁(t) + η(d-y(t))x₁(t), where y = w₁x₁ + w₂x₂ + b
• Step(1000): trains 1000 epochs
• Reset weights: resets weights to random values ranging [-0.5, 0.5]
• View