Cybiote

17 ms·

91.
by Cybiote 7y ago
Phantom limbs can cause/have pain, partly from the lack of expected predicted feedback. I cannot begin to imagine what sort of chaos, in the dynamical sense, a non-existent body would cause and how that would be represented in the part
92.
by Cybiote 8y ago
It's rare that a model introduced in a paper has fair comparisons. You'll want to see it used by many different authors before one can speak with confidence on claimed improvements. In this paper, Bellard makes a comparison to the
93.
by Cybiote 8y ago
These categorizations tend to be arbitrary and inconsistent because what is an intelligence is probably subjective. People consider kNN and naive bayes to be machine learning. One is "just" sorting and the other is just counting.
94.
by Cybiote 8y ago
Hi, what sort of complex derivations can you solve for with Sympy?
95.
by Cybiote 8y ago
I don't think that blog post is referring to the same paper. This new paper is about an experimental demonstration of Wigner's Friend: https://en.wikipedia.org/wiki/Wigner's_friend . Scott's blog pos
96.
by Cybiote 8y ago
Can we even say for certain it's an improvement over say something like TransformerXL? As far as I could see, the changes over GPT were a couple extra and tweak to layer normalizations, a small change to initialization and a change to
97.
by Cybiote 8y ago
Amazing results, how excited are you? :) You're right, I noted too that the comparison isn't direct but then, I wasn't justified in calling out the gap claim as wrong, so sorry for that. I think it'd be nice however, to
98.
by Cybiote 8y ago
Conscious processes are estimated to work on the order of 10^2 bits. Vision, at the retina, is estimated at 10^7 bits/sec. It drops another order of magnitude by V1. Also note that long as they're not isolated, a deaf and blind pe
99.
by Cybiote 8y ago
Agreed, the transformer's approach to memory/attention is certainly novel and a non-trivial departure from RNNs. It's also questionable the extent to which this is not over-fitting. Some are already finding phrases on the web
100.
by Cybiote 8y ago
It's becoming ever more certain that the transformer architecture is one of the largest contributions to AI (not merely machine learning, but AI), often beating LSTMs despite LSTMs being expressive enough to capture Turing Equivalence
101.
by Cybiote 8y ago
I do not think the divisions so sharp. Many formulations in RL can be seen as implicitly defining a particular (set of) differential equations. One can do something similar for certain simple GAs and write them as equations of an evolutiona
102.
by Cybiote 8y ago
Claude Shannon's Masters Thesis was on the application of Boolean Algebra to circuits, effectively founding digital circuit design. That should have been enough for anyone, but not for Shannon. His later work on Information theory has
103.
by Cybiote 8y ago
Type errors in functional languages (except Elm) are so unhelpful, you just get used to their general shape and your context to figure what the problem might be. Eventually these are rarely ever a problem but I agree that for beginners, typ
104.
by Cybiote 8y ago
There places where types will help when reading unfamiliar code. They'll also help when you're putting together more complex algorithms for novel architectures. They're less helpful if you're only stacking layers like le
105.
by Cybiote 8y ago
I don't dispute this. If you can remove memory transfer overhead and meet the requirements I mentioned, then GPUs will be much better. But there are many problems that do not fit that (SIMD per warp) regime. In short, GPUs are no panac
106.
by Cybiote 8y ago
My experience is that you need the additional caveat of a streaming fairly homogeneous and highly parallelizable approach or the memory transfer, branching and communication overheads will eat away nearly all the GPU gains. I've also n
107.
by Cybiote 8y ago
You've clarified your stance from nobody to nobody in business. That's good, although, I think that is opinion based on your experiences. I suspect that business will care if researchers can make it easy to learn on premise on the
108.
by Cybiote 8y ago
> Nobody cares how long it takes to train a model. This isn't true. It depends on your priorities and goals. Machine learning that spends most of its time unable to learn is not real AI. Some of us are interested in sample and energ
109.
by Cybiote 8y ago
This is excellent work. Is there a catch? Limited vocabulary versus open domain? Noise or accent sensitive? How large is the model file? What's the accuracy? It's amazing you were able to get anything beyond keyword spotting on su
110.
by Cybiote 8y ago
The name is not so bad as you two think. The act of measurement of a quantum state can be thought of as a filtering procedure after which a normalization operation is applied. This is the same protocol that occurs when conditioning on an ob
111.
by Cybiote 8y ago
Yes, it seems you're very much on the right track according to page 12, direct structure optimization: http://predictioncenter.org/casp13/doc/CASP13_Abstracts.pdf The document also answers some of my question
112.
by Cybiote 8y ago
The blog is light on details but it appears Alpha is being used for branding here. AlphaFold seems to depart in a significant way from the perfect information game playing agent's architecture. A few questions spring to mind: 1) What
113.
by Cybiote 8y ago
Indeed, I completely agree that some birds are very intelligent, managing to pack quite a lot of punch in such a small package. Parrots, corvids, even pigeons are examples of such intelligent birds. This is why I was careful to select Chick
114.
by Cybiote 8y ago
I agree with the general call for skepticism against the idea that displaced reference is so unique in the animal kingdom. Where I do not agree, is the implicit argument of "if a bee can do it then it's trivial". Within the a
115.
by Cybiote 8y ago
What I find most incredible about EURISKO is it did all that on what, by today's standards, is a very small amount of computational power. Just as outstanding is that, in all these years, in terms of open-ended problem solving, nothing
116.
by Cybiote 8y ago
This statement is fairly inaccurate. If you check Peter Madderom 1966 thesis, you'll see that it states the earliest work on automatic differentiation was done in the 1950s. It's just that back then, it was called Analyitc diffe
117.
by Cybiote 8y ago
Type annotations can be useful in helping a reader get a rough outline at a glance, of what some unfamiliar piece of code is doing. Especially under heavy use of broadcasting. The second is compactness is true only if you're chaining s
118.
by Cybiote 8y ago
Given how thoroughly the physicists who wrote the https://en.wikipedia.org/wiki/Frisch–Peierls_memorandum anticipated the use of nuclear weapons as deterrence, I do not think the supposition that physicists were unable
119.
by Cybiote 8y ago
The temperature of the sun at its surface is ~5772 Kelvin. To get power per unit area, use Stefan-Boltzman: σ * T^4, σ * (5772 K)^4 ≈ 6294 W/cm^2. Dividing the sun's luminosity (power) by its surface area will also give a similar
120.
by Cybiote 8y ago
Possible correction: this does not appear to be an example of machine bias. It's also important to keep in mind that there can be other sources (such as brittleness) of bad ML outcomes than bias. When I do an exact search for the Just

More ›