This is a technical project I made for the purpose of understanding the math behind neural networks. I finally figured it out!! explanations in the nodes + credits
Thanks to @MonkeyBean2 for his NNlib project, go check him out! it inspired the design of this project. Watch 3blue1brown for a full walkthrough of neural networks: https://www.youtube.com/watch?v=aircAruvnKk&list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi I don't think 3blue1brown explained error terms. To calculate the error terms, you first need to know the error terms of the outputs, which is (actual output - expected output) * derivative of the activation. Then, you go one layer back and calculate each of those error terms by adding up the products of each weight going out from the activation and the error term of the activation the weight goes into, all of this multiplied by the derivative of the activation. Repeat this process until you reach the input layer, hence the term "Backpropagation"
