TURBO MODE DOES NOT WORK!!! I don't feel like changing the code to make it work because it only takes like 3 minutes max to learn. D - Default Width N - Narrower Width Space - Pretrained (Shortcut!) (If trained on narrower width, can handle default width. If trained on default width, there's a very high chance it won't succeed on narrow width and will have to relearn.) #neural #network #ai #artificial #intelligence #trending #popular #featured #node #neuron #ml #machine #learning
Just click the green flag and watch it work it's magic! This is a very, very simple neural network. There are 4 neurons ("brain cells") in each AI. They have 2 input neurons: the x coordinate of the obstacle, and the x coordinate of their bodies. Each of those values are multiplied by different weights, then the values are added together. Then, a bias is added to that value (the bias is the 3rd neuron), resulting in the output neuron (the AI travels to the x coordinate corresponding with the output value)! The bias and weight values start as a random value, and as time goes on, it finds the weights and bias that allow it to succeed. When all 5 clones die, a new generation begins. Each of those new 5 clones' bias and weight values mutate slightly from the clone that survived the longest in the previous generation. Natural selection is pretty neat! This is the first neural network I have designed and created! I plan to translate the code to Java, next.
