-------Neuroevolution------- Use neural networks and genetic algorithms to solve problems! Any - change obstacle course Space - show variables Click the multiplier - change speed https://turbowarp.org/1093967392/fullscreen less lag -------Credits------- Everything is 100% my own! Huge thanks to @Dinosu , he helped me with some questions. I may add more in the future. Remix and add your own maps! For more on neuroevolution go check out Argonaut's vid: https://www.youtube.com/watch?v=9Zk_hY_CjiE -------Breakdown------- In essence, the AI population will improve gradually over time as a result of genetic algorithms. Each AI or "Agent" is equipped with three lines of sight and It's direction. It then plugs these numbers into its brain, or neural network. I won't go into detail, but each brain has 12 neurons: 4 input, 6 hidden and 2 output. Each neuron sums the weights*neuron and plugs it into an activation function to get a number.(Feed forward network). The Agent can move and rotate based of the two outputs. Based off the Agents performance, It will either reproduce or die (genetic algorithms). During reproduction, each weight(represented as an 8-bit string) will have a 1% chance of mutation to add variation in the given population. --------------------Hint------------------------------- https://scratch.mit.edu/studios/28715018/comments here is a link where you can propose projects to be featured :) *hint hint* Tags: #name #art #coding #triangles #squares ( ͡° ͜ʖ ͡°) Trig functions :P Cheating and not using parametric equations: x = cos(t) Y = sin(t) ^_____^ +10000000 aura for checking down here
