Neural Networks
Neurons, layers, backpropagation and architectures.
Neural Nets
Activation Functions: Why Neural Networks Need Nonlinearity
Activation functions add the nonlinearity neural networks need. Compare sigmoid, tanh, ReLU and GELU by output range and gradient behavior.
Neural NetsHow Backpropagation Trains Neural Networks, Step by Step
Backpropagation uses the chain rule to find how each weight affects the loss, then updates the weights. Follow one worked example with real numbers.