1,319 reputation
521
bio website suitdummy.blogspot.com
location Cambridge, MA
age
visits member for 2 years, 2 months
seen Mar 26 at 22:54

I once launched swi-prolog and asked it a question:

ely:~/home$ prolog
Welcome to SWI-Prolog (Multi-threaded, 64 bits, Version 5.10.1)
Copyright (c) 1990-2010 University of Amsterdam, VU Amsterdam
SWI-Prolog comes with ABSOLUTELY NO WARRANTY. This is free software,
and you are welcome to redistribute it under certain conditions.
Please visit http://www.swi-prolog.org for details.

For help, use ?- help(Topic). or ?- apropos(Word).

?- love(math) is unrequited.
true.

Feb
21
comment Optimization of the regularized least squares with gradient descent
You mean the analytic solution for ridge (Tikhonov regularized) regression? That seems to be what the question is looking for... not "usual least-squares" which is generally used to mean ordinary least squares.
Feb
19
answered what does “a distribution over distributions” mean?
Feb
2
awarded  Yearling
Feb
1
awarded  Tumbleweed
Jan
25
revised Aside from regression coefficients, what are commonly used approaches to measure one variable's “sensitivity” to another variable?
edited title
Jan
25
asked Aside from regression coefficients, what are commonly used approaches to measure one variable's “sensitivity” to another variable?
Dec
17
comment How do you “control” for a factor/variable?
Just to emphasize the point on this question (which is re-asked very often), it's good to consider that simply including a variable in a model is not guaranteed to "control" for its effect, even under extremely strong assumptions about the variable being monotonically related to the dependent variable. See the article that is linked in my other comment.
Dec
12
comment How to assess statistical significance of the accuracy of a classifier?
And some application domains, say financial markets, where you get to use the classifier in many many roughly independent cases, just being a bit better than chance (R-squared's of like 11% or 12% are considered great) can mean a lot. In those cases, if even the boosted classifier has R-squared of 15% that might be considered very good -- in which case it really matters if you can statistically resolve whether the weak classifiers are definitely better than guessing.
Dec
12
comment How to assess statistical significance of the accuracy of a classifier?
Not if you are boosting a bunch of weak classifiers, which is a very common activity. You may care about discrimination once you reach the fully boosted final classifier, but there's a lot of work between the start and the finish, and demonstrating that a complicated classifier empirically performs better than chance is important.
Dec
12
comment How to assess statistical significance of the accuracy of a classifier?
Classifiers that just barely beat random guessing can be extremely useful in some situations. Thus, having some test that quantifies confidence in a classifier being better than chance is also useful.
Dec
12
comment How to assess statistical significance of the accuracy of a classifier?
Classifiers that just barely beat random guessing can be extremely useful in some situations. Thus, having some test that quantifies confidence in a classifier being better than chance is also useful.
Dec
5
revised Measuring variation of one data set with respect to another data set
deleted 27 characters in body
Dec
5
answered Measuring variation of one data set with respect to another data set
Dec
5
awarded  Good Question
Dec
4
comment Does more variables mean tighter confidence intervals?
Those are two really important points. It's also worthwhile to mention how slight non-linearities can induce wrong signs on the coefficients, see for example the paper "Let's Put Garbage-can Regressions and Garbage-can Probits Where They Belong". Tremendous effort on the answer -- would be great to add clarifying remarks along these lines so that it's not misleading when people find this.
Nov
13
awarded  Popular Question
Oct
8
comment Machine learning on big data: capability of generalization
Some useful references: Zakharevich and Aaronson, particularly the second one and the section on look-up table arguments as they relate to the section on PAC learnability.
Oct
8
comment Machine learning on big data: capability of generalization
I severely disagree. See, e.g. Amdahl's Law. It's much more of a fundamental complexity issue, something that parallelism cannot solve. If something is poly-time on a distributed system, then it's poly-time on a serial system with a big constant out front. You do not gain super-poly-time speedup from parallel programming.
Oct
8
comment Machine learning on big data: capability of generalization
One may wish to have a computationally less expensive model. Observing all of the data might give you a giant look-up table, but the cost of performing the look-up for a given $x_i$ is too great -- you need a simpler, compressed heuristic that calculates $y_i$ reasonably well and finishes faster than the look-up would. While that is not the main problem of statistical learning theory, it is an important one.
Sep
18
comment Why does a 95% CI not imply a 95% chance of containing the mean?
They (the interested readers) may also wish to check out 0 and 1 are not probabilities.