Gneural Network - Tasks: task #14203, Implement DBN feedforward training...
You are not allowed to post comments on this tracker with your current authentication level.
task #14203: Implement DBN feedforward training for nnet.
Submitter: | Ray Dillinger <rayd> | ||
Submitted: | Sun 30 Oct 2016 05:16:15 PM UTC | ||
Should Start On: | Sun 30 Oct 2016 07:00:00 AM UTC | Should be Finished on: | Fri 30 Dec 2016 08:00:00 AM UTC |
Category: | None | Priority: | 6 |
Status: | None | Privacy: | Public |
Assigned to: | None | Percent Complete: | 0% |
Open/Closed: | Open | Effort: | 0.00 |
No files currently attached
Depends on the following items: None found
Items that depend on this one: None found
Carbon-Copy List
There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.
No changes have been made to this item
Powered by Savane 3.13-02a9.
Corresponding source code
Deep Belief Networks use forward propagation of error rather than backward propagation of error. Each node treats the difference between its actual activation level and the activation level it would have had if the next-layer nodes had activated it (rather than it activating them) as an error signal applying to the weights in the connections between them. Effectively the error is a bias in favor of the activated nodes reproducing the patterns in the activated nodes via the same weights that the activating nodes used to activate them.
This allows emergent pattern recognition starting at the input as opposed to result feedback starting at output. It is highly resistant to overfitting, and is essential to training deep networks. Further, it can be combined with backward propagation of error, facilitating a "meet in the middle" strategy.