ced

36 ms·

331.
by ced 12y ago
The article had very few data points, that's why it worked with 3 dimensions. The deciding factor is how N (effective number of data points) compares with p (effective number of features).
332.
by ced 12y ago
As a long-time Lisper in AI, Haskell does look intriguing, but I'm more worried about what I would lose. In particular, I'm curious about how DSLs fare in Haskell. I can write first-order logic sentences in Lisp, and they look and
333.
by ced 12y ago
David Mackay's Information Theory, Inference and Learning algorithms doesn't cover everything, but it ties up a lot of important topics. http://www.inference.phy.cam.ac.uk/mackay/itila/
334.
by ced 13y ago
I try to pass this stuff on to the next generation, but they are usually determined to make the same mistakes. Oh well. As someone working in the area of compilers... Do you have that advice written down anywhere?
335.
by ced 13y ago
John Bell himself wrote an entertaining paper railing on the orthodox perspective: Against Measurement. http://www.tau.ac.il/~quantum/Vaidman/IQM/BellAM.pdf
336.
by ced 13y ago
Ah, I'd always thought of EM as iterating over the parameters given a model, but I suppose it makes sense to see it as a search for a model, including parameters, that fits the data. Cool.
337.
by ced 13y ago
I'm fairly knowledgeable about Bayesian statistics, but I can't tell which technique you're referring to. It vaguely sounds like maximum likelihood, but ML doesn't iterate the model or priors. Could you explain further a
338.
by ced 13y ago
Off-topic: if an integral is an approximation of a physical process, then that process is an approximation of the integral. I.e. we could in principle have a "chip" running a miniature lab experiment (like water flowing) with sens
339.
by ced 13y ago
Common Lisp with SBCL: (defun double (x) (+ x x)) ; BigInts (or floats, for that matter) (defun double (x) (declare (type fixnum x)) (+ x x)) ; will emit an error if X is more than a long (defun double (x) (
340.
by ced 13y ago
Ahhh, I see what you mean, I think. Or maybe not. If I have 3 sample points, [1,1,1], then the standard deviation is 0, but where you take your Euclidean distance? If I'm not mistaken, with 3 sample points a, b and c that have an avera
341.
by ced 13y ago
Maybe I misunderstood what you meant, but I would say that the L_2 norm amounts to MAD. L2 means sqrt(x²+y²+...) and in 1D, that becomes sqrt(x²) = |x|
342.
by ced 13y ago
Are Futamura projections used anywhere, or are they mostly of theoretical interest?
343.
by ced 13y ago
Common Lisp has multiple return values and it's quite useful: (let x (f 12) ...) ; bind x to the first value returned by (f 12) and silently drop the others (let (x y) (f 12) ...) ; bind x to the first value and y to the seco
344.
by ced 13y ago
I think his point was about: There's probably lots of other things that can be overlooked rather than about machines and mice. It's a hard problem in science: how do you build a truly reproducible setup when you don't have
345.
by ced 13y ago
Model averaging can be quite useful when you're averaging over versions of the same model with different hyperparameters, e.g. the number of clusters in a mixture model. Yeah, but in this case, there's a crucial difference: with
346.
by ced 13y ago
I agree. As a Bayesian hoping to understand my data, P(X|M1) is useful: it's the probability I have for X under M1's modelling assumptions. Of course M1 is an approximation, but that's how science is done. You get to understa
347.
by ced 13y ago
There are standard examples where the minimal deterministic automaton for a language is exponentially larger than a minimal non-deterministic automaton Can anyone provide one such example, please?
348.
by ced 13y ago
For more context, I recommend looking at the two graphs here: https://en.wikipedia.org/wiki/Sunspot#Variation In particular, the Maunder minimum was a period of very low solar activity in the 17th century which contrib
349.
by ced 13y ago
If I remember correctly, the sunspot record is fairly accurate since we've had telescopes (~400 years ago), since they can last several months, and there aren't that many of them.
350.
by ced 13y ago
I don't know if overpopulation is a problem per se, but since pollution is roughly proportional to population, most of the environmental issues we face wouldn't be issues if we had 1/10 as many people.
351.
by ced 13y ago
every ODE that has one and only one solution has an exact solution Upvoted, but - is that a particular theorem, or are you just basically stating the definition of "having a solution"? Wouldn't your statement be true of any
352.
by ced 13y ago
That's terrible advice because people will drive instead. Or - more likely - they'll choose a closer destination for their summer vacation and avoid losing six days driving the round trip from New York to San Francisco.
353.
by ced 13y ago
Both the American Economic Review and the Review of Economic Studies rejected the paper for "triviality," while the reviewers for Journal of Political Economy rejected it as incorrect, arguing that if this paper was correct, then
354.
by ced 13y ago
Why is the Swiss system superior?
355.
by ced 13y ago
During the years we worked on Viaweb I read a lot of job descriptions. A new competitor seemed to emerge out of the woodwork every month or so. The first thing I would do, after checking to see if they had a live online demo, was look at t
356.
by ced 13y ago
Is the Japanese Python community really that big? What do they use it for?
357.
by ced 13y ago
I read it as intentionally ambiguous.
358.
by ced 13y ago
With Quicklisp, the situation has improved a bit. I've used tpapp's libraries without much of a problem. https://github.com/tpapp/
359.
by ced 13y ago
Disclaimer: I worked on the solar magnetic field for two summers. Whether the sun is chaotic in the technical sense (i.e. the butterfly effect) is still open to question [1]. The solar magnetic field is poorly understood. In the 1980s, ever
360.
by ced 13y ago
What makes APL different from, say, Lisp or Haskell? Do you have tutorials to recommend?

More ›