CODEWITHSUNDEEP
machine-learningneural-network
code muncher
code muncher

Reputation: 1682

Neural Network - updating weight matrix - back-propagation algorithm

I'm implementing neural network with the help of Prof Andrew Ng lectures or this, using figure 31 Algorithm.

I think I understood forward propagation and backward propagation fine, but confuse with updating weight (theta) after each iteration.

Q1. When and HOW to update weight (theta) matrix - theta1, theta2?

Q2. What is big Delta for? [Solved, thanks @xhudik]

Q3. do we have to add +1 (bias unit in input and hidden layer?)

Upvotes: 1

Views: 5711

Answers (1)

xhudik
xhudik

Reputation: 2444

Q1: is explained by @nikie (Kudos)

Q2: Andrew NG presentations are great. However, you are pointing to one which is high-level and want to understand details. What about this: http://galaxy.agh.edu.pl/~vlsi/AI/backp_t_en/backprop.html It gives you much more details with useful graphics

Upvotes: 6

Related Questions