One can easily build forward pass (forward propagation), for neurons or perceptrons. The learning process itself (which includes gradient descent) is something else entirely, I'd love to learn how to do that, I read a few articles and my stopping point always was multi-variable differential equations, I just couldn't go any further. You could obviously use training functions from existing libraries, but I'm not aware how to do it from scratch...
Comments
One can easily build forward pass (forward propagation), for neurons or perceptrons. The learning process itself (which includes gradient descent) is something else entirely, I'd love to learn how to do that, I read a few articles and my stopping point always was multi-variable differential equations, I just couldn't go any further. You could obviously use training functions from existing libraries, but I'm not aware how to do it from scratch...