Scratchnet is a fully customizable neural network made for others to use in their machine learning projects. The current example has a 2-3-2 neuron layout. It takes the mouse x and y position as input and is supposed to output the exact same values, basically meaning it's useless. I haven't programmed a bias neuron yet, but maybe it will be added in the future.
Current version: 1.4 All made by me using my own knowledge during a 14 hour car trip. This project is completely open source (all Scratch projects are in theory) but credits are always appreciated. See inside for more info. *OLD 1.3 INFO* The only problem with this machine learning project is that it stores the connection values as a list item and list items can only be 10240 characters long. This seems like a lot, but it's not even enough for a 30-50-20 network. @TheLogFather found a way around this (https://scratch.mit.edu/projects/24934494/) so I'll try to implement that in 1.4.
