[HN Gopher] GHz spiking neuromorphic photonic chip with in-situ ...
___________________________________________________________________
GHz spiking neuromorphic photonic chip with in-situ training
Author : juanviera23
Score : 109 points
Date : 2025-08-04 11:21 UTC (11 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| rf15 wrote:
| Appreciating that not everyone tries to optimise for LLMs and we
| are still doing things like this. If you're looking at HN alone,
| it sometimes feels like the hype could drown out everything else.
| danielbln wrote:
| There is massive hype, no doubt about it, but lets also not
| forget how LLMs have basically solved NLP, are a step change in
| many dimensions and are disrupting and changing things like
| software engineering like nothing else before it.
|
| So I hear you, but on the flip side we _should_ be reading a
| lot about LLMs here, as they have a direct impact on the work
| that most of us do.
|
| That said, seeing other papers pop up that are not related to
| transformer based networks is appreciated.
| larodi wrote:
| Thank you, brother. Besides not all that goes in HN is
| strictly LLM, really dunno why the scare.
| karanveer wrote:
| I couldnt agree more.
| msgodel wrote:
| It's just a single linear layer and it's not clear to me that the
| technology is capable of anything more. If I'm reading it
| correctly it sounds like running the model forward couldn't even
| use the technology, they had to record the weights and do it the
| old fashion way.
| roflmaostc wrote:
| Would you have discredited early AI work because they could
| only train and compute a couple of weights?
|
| This is about first prototypes and scaling is often easier than
| the basic principle.
| msgodel wrote:
| Is this actually capable of propagating the gradient and
| training more complex layers though?
|
| A lot of these novel AI accelerators run into problems like
| that because they're not capable of general purpose
| computing. A good example of that are the boltzman machines
| on Dwave's stuff. Yeah it can do that but it can only do that
| because the machine is only capable of doing QUBO.
| roflmaostc wrote:
| For inference we do not care about training, right?
|
| But if we could make cheaper inference machines available,
| everyone would profit. Isn't it that LLMs use more energy
| in inference than training these days?
| fjfaase wrote:
| Nice that they can do the processing in the GHz range, but from
| some pictures in the paper, it seems the system has only 60
| 'cells', which is rather low compared to the number of cells
| found in brains of animals that display complex behavior. To me
| it seems this is an optimization in the wrong dimension.
| _jab wrote:
| I suspect practicality is not the goal here, but rather a proof
| of concept. Perhaps they saw speed as an important technical
| barrier to cross
| khalic wrote:
| A lot of unrigorous claims for an abstract...
| kadushka wrote:
| Maybe try simulating the algorithms in software before building
| hardware? People have been trying to get spiking networks to work
| for several decades now, with zero success. If it does not work
| in software, it's not going to work in hardware.
| vessenes wrote:
| This seems to work in hardware, per the paper. At least to 80%
| accuracy.
| good_stuffs wrote:
| >If it does not work in software, it's not going to work in
| hardware.
|
| Aren't there limits to what can be simulated in software?
| Analog systems dealing with infinite precision, and having
| large numbers of connections between neurons is bound to hit
| the von Neumann bottleneck for classical computers where memory
| and compute are separate?
| juliangamble wrote:
| "Zero success" seems a bit strong. People have been able to get
| 96% accuracy on MINST digits on their local machine.
| https://norse.github.io/notebooks/mnist_classifiers.html I
| think it may be more accurate to say "1970s level neural net
| performance". The evidence suggests it is a nascent field of
| research.
| cwmoore wrote:
| Retina-inspired video recognition using light. Cool. May be a
| visual cortex next year.
| vessenes wrote:
| Ghz speed video processing, even if we only get very basic
| segmentation or recognition out of it, is probably crazy useful.
| Need to face recognize every seat at a stadium?
|
| Well, if you have enough cameras, 60,000 seats could be scanned
| 250 thousand times a second. Or if you want to scan a second of
| video at 60fps, you'd be able to check all of them at a mere 4
| thousand times a second.
|
| Anyway, good to see interesting raw research. I imagine there are
| a number of military and security use cases here that could fund
| something to market (at least a small initial market).
___________________________________________________________________
(page generated 2025-08-04 23:01 UTC)