Blog topic

How machines learn

The ideas under every learning method: the loss, the gradient, one step downhill, and what goes wrong when the step is too large or the model too flexible.

Articles