annotate tvii/gradient_descent.py @ 56:5867e4a10fae

[doc] link to deeplearn.js
author Jeff Hammel <k0scist@gmail.com>
date Fri, 29 Sep 2017 10:00:31 -0700
parents 857a606783e1
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
44
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
1 """
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
2 gradient descent for neural networks
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
3 """
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
4
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
5 # Each loop:
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
6 # - computes predictions
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
7 # - compute derivatives: dw, db
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
8 # update: w = w - alpha * dw; b = b - ....
857a606783e1 [documentation] notes + stubs on gradient descent
Jeff Hammel <k0scist@gmail.com>
parents:
diff changeset
9