6
$\begingroup$

Has anybody read or have access to

Alex Andrew Significance Feedback in Neural Nets Report of Biological Computer Laboratory, University of Illinois, Urbana, IL GM-10718-03 TR No 5 September 1965 pages 1--12

and can confirm the contribution this made to backprop?

Now, the basic routine is done by economists since the late 19th century for inferring value to factors of factors of ... production, with the chain rule, but for neural networks, this is cited by cybernetics oriented persons as the very first application. Is it? Margaret Boden cites Paul Werbos as first but that I have read (it was republished in his book) but Werbos's dissertation came a decade after Andrew's report.

I have seen this cited as the first modern backpropagation in cybernetics paper, and not surprisingly it is from the place where McCulloch, Ashby, et al worked. But I can't access it to confirm. Alexander passed away recently so can't contact him either. Anybody have this thing, or read it, or have it scanned? Archive.org has some of the other reports (for example, on asynchronous computers) but not this one ...

UPDATE: Interlibrary loan cannot get ahold of a copy of the first edition. They gave up on all requests! Does anybody have this paper, or the reprint of it from the 1990s?

Improve this question
$\endgroup$
3
  • $\begingroup$ Presumably you've tried the University of Illinois, Urbana's library? $\endgroup$
    – IanF1
    Jul 14, 2016 at 6:03
  • $\begingroup$ Yes, have a loan request there, no response yet from UoI to my university library... $\endgroup$
    – Gottfried William
    Jul 16, 2016 at 14:06
  • 1
    $\begingroup$ I found a scan of it online at this link: bclarchive.net/browse (there doesn't seem to be a way to link directly to the document, but it's at the top of the page if you sort by fiche). I would post this as an answer but it would be link-only since I can't comment on the document's contributions (I'm not very familiar with neural nets). $\endgroup$
    – texdr.aft
    Nov 2, 2020 at 2:21

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.