TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Analog-Cartpole: A hybrid analog/digital reinforcement learning experiment

28 pointsby ghosthamletabout 5 years ago

3 comments

rough-seaabout 5 years ago
Neural networks are largely operating on 32 or 64 bit floating point numbers. Many millions of them. If you think how this works at the hardware level there are many little cores in a GPU for operating on matrices of these numbers. To feed a 32 bit float into one of these, it requires 32 wires - deep in the silicon.<p>Also note that neural networks are inherently noisy. We often insert a bit of noise into various parts of the computation graph.<p>In analog circuity only a single wire (or maybe two if you’re using a differential pair) would be needed to represent a noisy float.<p>If we had some sort of IC that could dynamically configure large analog computations, it may allow NN compute-graph computations to be improved by orders of magnitude. 1 wire instead of 32. Real noise instead of artificial.<p>Have people ever tried to something like an analog FPGA?
评论 #22605310 未加载
ytersabout 5 years ago
I don&#x27;t get the interest in analog. Whatever the secret sauce that makes Real Intelligence (TM), analog is not it.
ganzuulabout 5 years ago
This might be useful for reducing the &quot;reality gap&quot; in robot simulation.