A Neural Network I made expanding on one I made some time ago, except this AI can learn five different calculations! I've also added some general improvements to other parts of the project :) --Check out my other projects, I've made several AIs, videogames, and more! -- Depending on the operation, it can take anywhere from 50,000 to over 100,000 iterations to get less than ten percent error, demonstrating the quantity over quality approach of this single-dimensional weight adjustment (because I'm not building multi-var calculus into Scratch, sorry) and the low number of nodes involved.
Speed Problems? - Once the list of errors starts to accumulate above 45,000, it may start to slow down the calculations; press "p" to purge the list. - The neural network may work slowly due to the small number of nodes and Scratch's processing issues - have patience and press space to do more training each time it finishes a 15,000 iteration set! Make sure to like and subscribe lol Thanks to https://scratch.mit.edu/projects/196949694/#player and Wikipedia #Neural #Network #AI #Artificial #Neural_Network #Dab
