Dr. Mark Humphrys

School of Computing. Dublin City University.

Home      Blog      Teaching      Research      Contact

My big idea: Ancient Brain


CA114      CA170

CA668      CA669      Projects

The Back-propagation algorithm

The back-prop algorithm for supervised learning from exemplars is:

  1. Send in inputs Ii for this exemplar.
  2. Calculate outputs yk
  3. Get given correct outputs Ok
  4. Measure E.

  5. wjk weights:
    1. Calculate all the   =   yk ( 1 - yk ) ( yk - Ok )
    2. Calculate all the   =     yj
    3. For all j,k:

  6. wij weights:
    1. Calculate all the   =   yj ( 1 - yj )   Σ k ( wjk )
    2. Calculate all the   =   Ii
    3. For all i,j:

  7. Repeat (next exemplar).

See sample code.

ancientbrain.com      w2mind.org      humphrysfamilytree.com

On the Internet since 1987.

Wikipedia: Sometimes I link to Wikipedia. I have written something In defence of Wikipedia. It is often a useful starting point but you cannot trust it. Linking to it is like linking to a Google search. A starting point, not a destination. I automatically highlight in red all links to Wikipedia and Google search and other possibly-unreliable user-generated content.