Connectionists: Request for info related to hardware for perceptrons

[Erik Winfree]

This New York Times article is quite the blast from the past:

https://www.nytimes.com/1958/07/13/archives/electronic-brain-teaches-itself.html

Not quite the serious discussion you’re looking for, but still.  Ever-so-slightly more grounded:

https://news.cornell.edu/stories/2019/09/professors-perceptron-paved-way-ai-60-years-too-soon

My dad, Art Winfree, was one of those Cornell undergraduates helping him build the electronic perceptron machines.  He told me in 2001 that: 

Spkg of antiquity, Nature 29 March reminisces over "bionics", as it was called around 1960 when it lured me (highschool senior) in to engineering and biology.  First stint of real research was in Cornell Aeronautical Labs where Frank Rosenblatt, my mentor and idol, was building "Perceptrons". Main problem was the synapse: how to store cumulative experience? Tried capacitors, but they leaked overnight. Finally an electrochemical cell involving silver deposition, sorta like a battery in reverse, with cumulatively altered resistance, modified by Hebb rule. Then Carver Mead started along silicon lines and eventually came up with a way to store electrons a long time w/o leaking. 

Wikipedia describes the machine that perhaps you are referring to as the "Mark 1 Perceptron” and has a picture (https://en.wikipedia.org/wiki/Perceptron), but refers to Bishop (2006) for details (also see http://csis.pace.edu/~ctappert/srd2011/rosenblatt-contributions.htm).  The Mark 1 is referred to in Rosenblatt’s 600-page “Principles of Neurodynamics” (1961), and has a low-quality photo (https://apps.dtic.mil/sti/pdfs/AD0256582.pdf), but implementation details are sparse, or at least, I could not easily find them.

There must be a better description somewhere.

Best regards,
Erik

> On Dec 18, 2021, at 3:59 AM, Prof Leslie Smith <l.s.smith at cs.stir.ac.uk> wrote:
> 
> Dear all:
> 
> I'm writing a paper on hardware for Neural networks, and I recollect that
> in the 1960's there was a discrete-transistor-sized adaptive resistor
> developed for perceptrons, possibly by Rosenblatt. I can't find any record
> of it, or anything in USPTO: can anyone supply me with a link or a paper
> on it?
> 
> with much thanks (and apologies too: I can't currently get into my office
> to look at my old papers)
> 
> --Leslie Smith
> 
> -- 
> Prof Leslie Smith (Emeritus)
> Computing Science & Mathematics,
> University of Stirling, Stirling FK9 4LA
> Scotland, UK
> Tel +44 1786 467435
> Web: http://www.cs.stir.ac.uk/~lss
> Blog: http://lestheprof.com
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20211220/2ec99c89/attachment.html>