[FRIAM] NickC channels DaveW
Steve Smith
sasmyth at swcp.com
Wed Jan 18 13:09:47 EST 2023
On 1/18/23 9:30 AM, cody dooderson wrote:
> Would it be possible for some hologram people (maybe Steve Smith and
> associates), to turn one of these convolutional neural networks into
> an optical convolutional network?
Nevermore!
Sadly, even when Fred Unterseher (LAVA Holographics) was still alive and
had his lab functioning, I don't know that we could have pulled more
than a mundane demonstration of even the simplest example of this off.
There were times when we tried to put some formal understanding on the
very ad-hoc/intuitive *mandalas* he created as an
art/spiritual/meditative form. There were times when two or more of
them hanging in a window (so very nearly perfect coherent light from the
sun) would throw diffraction patterns back and forth and we were left
speculating as to whether they were perhaps doing some kind of
interesting (if not directedly useful) "computation".
> I could see a bunch of layers of tiny lenses with different 'weights'
> doing the same thing as say a YOLO object detection network.
I had a short side-gig with the State Department right after the fall of
the Soviet Union when they were actively running a "keep a Soviet
Scientist off the Street" program to augment the "loose Nuke" wrangling
going on at the time. One of the more fascinating papers I reviewed
for them was on "optical computing" and at the time (I was about your
(Cody's) age at the time, maybe a few years younger even) felt to be
very close to what you are talking about.
Standing in for your term "tiny optical lenses" would have been
"holographic optical elements" (HOEs) which in fact might not have been
quite as *discrete* as lenses, but rather a (near)
continuum/distribution of diffraction plenum (my made up term). Fred's
specialty was in using organic light-sensitive materials which recorded
constructive/destructive patterns as thickness in a gel rather than the
silver-halide photographic processes which recorded as silver-grain
sizes. The optics/physics implications are subtle and I don't pretend
to be able to dredge much of *that* out of my rear-view mirror but I
believed (with limited support) at the time that the differences *might*
have been important (especially in the Mandala-Mandala-Mandala
interactions) to the fidelity of the wave-structures being iterated
optically.
Thanks for the nod, even if all I can offer in response is some
neurodivergent spastic head (and tongue) wagging back at ya!
- Steve
>
> _ Cody Smith _
> cody at simtable.com
>
> * YOLOv7 https://arxiv.org/pdf/2207.02696.pdf
>
>
> On Wed, Jan 18, 2023 at 8:06 AM Marcus Daniels <marcus at snoutfarm.com>
> wrote:
>
> Cerebras says they can scale past 1 trillion parameters..
>
> Cerebras-MemoryX-SwarmX-CS-2_banner.png
> Wafer Scale to ‘Brain-Scale’ – Cerebras Touts Linear Scaling up to
> 192 CS-2 Systems
> <https://www.hpcwire.com/2021/08/24/wafer-scale-to-brain-scale-cerebras-touts-linear-scaling-for-up-to-192-cs-2-systems/>
> hpcwire.com
> <https://www.hpcwire.com/2021/08/24/wafer-scale-to-brain-scale-cerebras-touts-linear-scaling-for-up-to-192-cs-2-systems/>
>
> <https://www.hpcwire.com/2021/08/24/wafer-scale-to-brain-scale-cerebras-touts-linear-scaling-for-up-to-192-cs-2-systems/>
>
> That would be a power budget of a HPC center, but not out of the
> ordinary. Less than 10 MW. AWS, Azure, Google and national labs
> have facilities like that.
>
>> On Jan 18, 2023, at 12:41 AM, Pieter Steenekamp
>> <pieters at randcontrols.co.za> wrote:
>>
>>
>> I totally agree that realizable behavior is what matters.
>>
>> The elephant in the room is whether AI (and robotics of course)
>> will (not to replace but to) be able to do better than humans in
>> all respects, including come up with creative solutions to not
>> only the world's most pressing problems but also small creative
>> things like writing poems, and then to do the mental and physical
>> tasks required to provide goods and services to all in the world,
>>
>> Sam Altman said there are two things that will shape our future;
>> intelligence and energy. If we have real abundant intelligence
>> and energy, the world will be very different indeed.
>>
>> To quote Sam Altmen at
>> https://economictimes.indiatimes.com/tech/startups/intelligence-energy-sam-altmans-technology-predictions-for-2020s/articleshow/86088731.cms
>> :
>>
>> "intelligence and energy have been the fundamental limiters
>> towards most things we want. A future where these are not the
>> limiting reagents will be radically different, and can be
>> amazingly better."
>>
>>
>>
>> On Wed, 18 Jan 2023 at 03:06, Marcus Daniels
>> <marcus at snoutfarm.com> wrote:
>>
>> Definitions are all fine and good, but realizable behavior is
>> what matters. Analog computers will have imperfect
>> behavior, and there will be leakage between components. A
>> large network of transistors or neurons are sufficiently
>> similar for my purposes. The unrolling would be inside a
>> skull, so somewhat isolated from interference.
>>
>> -----Original Message-----
>> From: Friam <friam-bounces at redfish.com> On Behalf Of glen
>> Sent: Tuesday, January 17, 2023 2:11 PM
>> To: friam at redfish.com
>> Subject: Re: [FRIAM] NickC channels DaveW
>>
>> I don't quite grok that. A crisp definition of recursion
>> implies no interaction with the outside world, right? If you
>> can tolerate the ambiguity in that statement, the artifacts
>> laying about from an unrolled recursion might be seen and
>> used by outsiders. That's not to say a trespasser can't have
>> some sophisticated intrusion technique. But unrolled seems
>> more "open" to family, friends, and the occasional acquaintance.
>>
>> On 1/17/23 13:37, Marcus Daniels wrote:
>> > I probably didn't pay enough attention to the thread some
>> time ago on serialization, but to me recursion is hard to
>> distinguish from an unrolling of recursion.
>>
>> --
>> ꙮ Mɥǝu ǝlǝdɥɐuʇs ɟᴉƃɥʇ' ʇɥǝ ƃɹɐss snɟɟǝɹs˙ ꙮ
>>
>> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. ---
>> -.. .
>> FRIAM Applied Complexity Group listserv
>> Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p
>> Zoom https://bit.ly/virtualfriam
>> to (un)subscribe
>> http://redfish.com/mailman/listinfo/friam_redfish.com
>> FRIAM-COMIC http://friam-comic.blogspot.com/
>> archives: 5/2017 thru present
>> https://redfish.com/pipermail/friam_redfish.com/
>> 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
>> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. ---
>> -.. .
>> FRIAM Applied Complexity Group listserv
>> Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p
>> Zoom https://bit.ly/virtualfriam
>> to (un)subscribe
>> http://redfish.com/mailman/listinfo/friam_redfish.com
>> FRIAM-COMIC http://friam-comic.blogspot.com/
>> archives: 5/2017 thru present
>> https://redfish.com/pipermail/friam_redfish.com/
>> 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
>>
>> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
>> FRIAM Applied Complexity Group listserv
>> Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom
>> https://bit.ly/virtualfriam
>> to (un)subscribe
>> http://redfish.com/mailman/listinfo/friam_redfish.com
>> FRIAM-COMIC http://friam-comic.blogspot.com/
>> archives: 5/2017 thru present
>> https://redfish.com/pipermail/friam_redfish.com/
>> 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
> FRIAM Applied Complexity Group listserv
> Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom
> https://bit.ly/virtualfriam
> to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
> FRIAM-COMIC http://friam-comic.blogspot.com/
> archives: 5/2017 thru present
> https://redfish.com/pipermail/friam_redfish.com/
> 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
>
>
> -. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
> FRIAM Applied Complexity Group listserv
> Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoomhttps://bit.ly/virtualfriam
> to (un)subscribehttp://redfish.com/mailman/listinfo/friam_redfish.com
> FRIAM-COMIChttp://friam-comic.blogspot.com/
> archives: 5/2017 thru presenthttps://redfish.com/pipermail/friam_redfish.com/
> 1/2003 thru 6/2021http://friam.383.s1.nabble.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20230118/31782a7a/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Cerebras-MemoryX-SwarmX-CS-2_banner.png
Type: image/png
Size: 214215 bytes
Desc: not available
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20230118/31782a7a/attachment-0001.png>
More information about the Friam
mailing list