Neural Networks w/ JAVA (Prototype Project 06) - Solve XOR w/ Hill Climbing
3,456 views · Published 24 May 2017 · 23:24 · Indexed 20 September 2026
Channel: Prototype Project · 2017 · Education
► get access to members-only video contents + support:
https://www.youtube.com/channel/UCUvwlMMaeppKPdtAK8PxO8Q/join
► download ai source code @ https://sites.fastspring.com/prototypeprj/instant/ai
► download crypto source code @ https://sites.fastspring.com/prototypeprj/instant/crypto
00:06 try running 'neural network tutorial 03' w/ XOR training data
00:56 not possible to have one line separating the 0s and 1s for XOR trainig data
01:33 demo prebuilt version of the application
02:32 use a neural network with 5 neurons (2 in input layer, 2 in hidden layer, and 1 in output layer), and where each neuron has 2 inputs and a threshold
03:16 proceed when 'adjusted error' is smaller or equal to current error
04:38 code the application
04:57 go over the various classes in this application
07:04 use XOR as the training data
07:19 log sigmoid transfer function in Neuron class
08:28 code NeuralNetwork class
11:00 calculate root mean square error
12:10 code run method of NeuralNetwork class
15:38 code class containing 'Hill Climbing' logic
20:30 code Driver class
22:38 test run completed application
