101–110 of 131 posts
Dynamic Routing Between Capsules the long awaited paper from Geoffrey Hinton (although Sara Sabour is principal author, Hinton has been hinting at this for a while now) on an alternative to back-propagation https://arxiv.org/abs/1710.09829
Unfortunately, it does not look very promising so far.
What's beautiful about this paper is that it shows a provable separation in the power of quantum circuits over classical circuits. That is in contrast to many cases where quantum advantage is only suspected (the most notorious being Shor's algorithm: Factoring could be in P).
Another cool thing is that this separation is fundamentally connected to quantum entanglement and the challenge of simulating measurements on multipartite quantum states with shared randomness and classical communication. This gives me some hope that this points to a path for future deep insights separating quantum from classical computing.
Earlier quoted context omitted.
Link to paper that cedes control of Borg?
Blog post: https://deepmind.com/blog/deepmind-ai-reduces-google-data-ce... Whitepaper: https://docs.google.com/a/google.com/viewer?url=www.google.c...
Leslie Lamport's 1978 "State the Problem Before Describing the Solution" [0]. On his web page the author adds that "The title says it all. This one-page note is as relevant today as when I wrote it. Replace "describing the solution" by "writing the program" and it becomes a practical recipe for improving software." Herewith, the paper in full: "After several years of writing papers in computer science, I discovered t…
Indeed. Many -- perhaps even a majority of -- startup landing pages break this rule.
Does anyone have any suggestions on finding CS papers worth reading beyond those papers listed here?
The author presents a very approachable technique for building pedagogical compilers by starting with a tiny subset of the language - a language that can accept integers and print it and incrementally grows it into Scheme. Every step yields a fully working compiler for a progressively expanding subset of Scheme. Every compiler step produces real assembly code that can be assembled and executed directly by the hardware.
I've dabbled with compilers for a while now and in my experience I found this to be among the best ways to get people excited about compilers. There is some code available from Nada Amin[2] for the curious and I'm re implementing it from scratch again[3].
1. http://scheme2006.cs.uchicago.edu/11-ghuloum.pdf
I highly recommend people give this paper a read. I think it points the way to a radical redesign of fundamental parts of the system stack over the next 5-10 years. If you work in systems and you aren’t thinking about this stuff, you’re about to be lapped. The Case for Learned Index Structures: https://arxiv.org/pdf/1712.01208v1.pdf
Yo! This is what I'm talkin' about! Replace directories with self-consistent neural nwots [neural blackbox thingie of weights and activation vals]. It's cool, but for every convenience we take in the direction HD space we backpeddle in processing [meaning, although the dir structure could be super condensed this way with a learned index, this approach is less flexible to cold or not-cpu-related analysis]. It's very c…
Can you expand on this bit? I'm not sure I understand what you mean
Earlier quoted context omitted.
I'm gonna throw some cold water on this and say this is not a new paradigm by any means. https://www.sigarch.org/the-unreasonable-ineffectiveness-of-... However, it's certainly possible that the time for this idea has come. Google is probably in the best position to apply it. I will say that after having worked at Google for over a decade, some of it on data center performance, there are plenty of inefficiencies that…
The article you link ends on a pretty weak claim, math didn't obsolete biologists and ML won't obsolete systems folks, but every time you write a heuristic you would probably get better results from a model.
https://research.google.com/pubs/pub43146.html
This is from the team at Google with the longest experience deploying machine learning (early 2000's, pre-deep learning).
(Contrary to popular belief, machine learning was basically unused in Google's search ranking until pretty recently.)
Earlier quoted context omitted.
I'm gonna throw some cold water on this and say this is not a new paradigm by any means. https://www.sigarch.org/the-unreasonable-ineffectiveness-of-... However, it's certainly possible that the time for this idea has come. Google is probably in the best position to apply it. I will say that after having worked at Google for over a decade, some of it on data center performance, there are plenty of inefficiencies that…
Have you seen the results when they let a trained model manage Borg? The power reductions were immediate, non-trivial, and performance stayed the same. There's your scheduling result for you. Look at it this way. As the paper points out, a Hashtable is just a heuristic that works fairly well in the worst case and reasonably well in the average case. No one would argue that you couldn't hand-roll an algorithm that is…
Also, predicting power usage is a different problem than scheduling. It's not "managing Borg" by any stretch of the imagination. I'm not saying that's impossible, but it's not what was claimed.
Also, what did Deep Blue lead to? I'm not saying it wasn't impressive, but it didn't generalize AFAIK. You are making a huge leap and claiming that a technique that isn't even deployed generalizes.
What did Watson lead to?
https://www.technologyreview.com/s/607965/a-reality-check-fo...
https://respectfulinsolence.com/2017/09/18/ibm-watson-not-li...