News for hackers. Stuff that matters.
Re:Thermodynamic Natural Gradient Descent by kaelan123 on Friday May 24, 2024 @02:05PM (#40468807)
First author of the paper here. That's it indeed! One thing is that this is entirely CMOS-compatible. You could also do something similar with optics or other platforms, but we chose electronic circuits for this reason specifically.
Re:Thermodynamic Natural Gradient Descent by CamperBob2 on Friday May 24, 2024 @02:20PM (#40468981)
By that remark I meant "digital CMOS," in the sense of elements that store state information discretely in flip-flops or gate insulators rather than continuously with analog integrators.

Very cool work in any event, though! Best of luck with the ongoing R&D.

Re:Thermodynamic Natural Gradient Descent by anigbrowl on Friday May 24, 2024 @10:34PM (#40472219)
Can you implement this on an FPGA?
Re:Thermodynamic Natural Gradient Descent by simne on Saturday May 25, 2024 @11:16AM (#40475521)
Any good electronics engineer can. Question is just who will pay for it? Or closer to YC environment, question is monetization.