I have been searching for more than one week which learning methods were used in Neurogrid.

But I only found descriptions of its architecture (chips, circuits, analog and/or digital components, performance results), everything but no clue on how it updates the weights.

In my opinion, I think that it cannot be gradient descent (with back-propagation), as the topology of the neurons in a chip, for example in the neurocore of Neurogrid, is a mesh or grid.

Do you know where I could find this kind of information?

One neurocore of Neurogrid

  • $\begingroup$ This is a relatively new field. I got a feel of it by reading a lot of papers. You should read the papers by the authors in this field, as far as I know no authorative sources exist except some collection of chapters by famous author in this field. I can't recall how I found it but I think you'll find it by a little but of digging around. $\endgroup$ – DuttaA Jun 27 '19 at 15:55
  • $\begingroup$ Hey thanks a lot for answering, could you please give me these names or articles ? I looked for those for hours and hours without any result. $\endgroup$ – ladangvu Jun 27 '19 at 18:10
  • $\begingroup$ sorry it was quite some time back..All I can say is pick a famous researcher..I think a few of them are in uni of Manchester? Or Stanford? Or eth Zurich? Then check their papers on these topics and you'll see in the learning algo they'll probably cite some paper and like this follow the paper trail, but the downside is that these are quite advanced topics and I think some papers assume you have significant background knowledge. $\endgroup$ – DuttaA Jun 27 '19 at 18:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.