Build your own Neural Network – with PHP!
summary
Builds a feed‑forward ANN from scratch in PHP that actually learns the XOR function using a 2-2-1 topology, sigmoid activation, mean squared error and gradient‑descent backpropagation. Code uses the math‑php matrix library and mentions PHP‑SCI for faster array ops. Initializes weights randomly, biases zero, uses learning rate 0.1, trains on four labeled examples until error <0.001 (stops around ~10,000 epochs), and returns predictions near 0 or 1, so the implementation works. Source code is on GitHub. XOR dataset math-php
AI-written summary · about
chapters (17)
- 0:00Introduction
- 1:18Why this talk
- 5:15Neuron
- 6:48Learn about parameters
- 9:16What is machine learning
- 13:49XOR
- 16:30Topology
- 16:49Parameters
- 19:50Activation
- 21:58Initialization
- 25:05Forward Propagation
- 27:12Cost Function
- 28:39Back Propagation
- 32:37Update Parameters
- 34:35Recap
- 37:43Challenge
- 41:26Resources
related talks
48mBuild your own Neural Network, with PHP!
44mMachine Learning with Artificial Neural Networks in Clojure
51mMachine Learning Exposed Deep and Reinforcement Learning
43m(Deep) Learn You a Neural Net for Great Good!
44mMachine Learning Exposed
57mWriting a Neural Net from Scratch
29mLearn Neural Networks With Go, Not Math
01h 25mBuilding & Optimization of Neural Networks from Scratch with Numpy