Скачать с ютуб Backpropagation Done Right! Caclulus for ML - Part Many/Many в хорошем качестве

Backpropagation Done Right! Caclulus for ML - Part Many/Many

In this lecture, we use the knowledge we developed from calculus to derive the mathematics of backpropagation from scratch. This is the power horse of AI! #ai #machinelearning #reinforcementlearning #deeplearning 00:00 What is Backprop 03:45 Loss Function and Partial Derivatives 07:55 Example K-Layer MLP 16:31 Gradient on K-1 Layer 22:12 Gradient of K-2 Layer - GPT4s Rhythm: In neural networks where dreams are shaped, backpropagation helps escape, Mistakes of yore, now insights pour, as layers adapt and reshape. Through hidden layers, errors flow, as gradients teach weights to grow, In reverse they tread, refining the spread, till perfect predictions show.