#gradient-descent

2 items tagged “gradient-descent”

TILs (2)

Why cross-entropy, and what −1/m is really doing

While building a single neuron from scratch in NumPy for the DLH machine-learning curriculum (classifying MNIST handwritten digits), I had to answer the question every ML beginner meets: how do you sc…

Gradient descent is just walking downhill — and why (A − Y) is enough

While building a single neuron from scratch in NumPy for the DLH machine-learning curriculum (classifying MNIST handwritten digits), I kept hearing "gradient descent" as the magic that makes models le…