Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
alextp
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
24 ms
·
181.
▲
by
alextp
16y ago
Those algorithms are actually surprisingly simple to code most of the time, if you're using a high-level language and aren't worried about scaling. Scaling them is hard, but for that you can already find good, scalable implementations out t
182.
▲
by
alextp
16y ago
I think most of this is due to the fact that most ML techniques need lots of specific knowledge to apply correctly. Many introductions focus too much on specific algorithms, for example, while ensemble methods are probably best in real-worl
183.
▲
by
alextp
16y ago
Actually, the best recent results in this area involve discriminatively trained stacked auto-encoders using second-order algorithms (for which the code, as of Sept 2010, is still private to the group at the university of toronto), as descri
184.
▲
by
alextp
16y ago
This is true, but the very good deep learning results are very recent (from less than 5 years ago), and they generally require very different models and/or training strategies from the standard multilayer perceptron you find in most AI text
185.
▲
Good, freely-available textbooks in machine learning
(metaoptimize.com)
133 points
by
alextp
16y ago
|
33 comments
186.
▲
by
alextp
16y ago
A sentence I extracted from a particularly idiomatic piece of portuguese: > O primeiro, presença ofuscante na consciência social, tem nome definido e inscrição temporal supostamente clara: a crise financeira global. If you translate it
187.
▲
by
alextp
16y ago
Not endless. I'm brazillian, and more than once when translating a good, terse, phrase in english to portuguese I felt the need to "pad it up" with rethorical and discourse markers that have no meaning in themselves; conversely, when transl
188.
▲
by
alextp
16y ago
I'm brazillian and can't speak (but can read) spanish. The translations of bolaño to english I've glanced over read really bad, when compared to the original (or the translations to portuguese). I think terseness is an issue here.
189.
▲
by
alextp
16y ago
I still haven't been able to finish Infinite Jest. It's just... not that interesting, I guess. Part of me thinks English is not a very good language for what's been attempted there; English as it is usually written is too terse and direct,
190.
▲
by
alextp
16y ago
There is a fantastic documentary about a documentary, Santiago, ( http://pt.wikipedia.org/wiki/Santiago_(document%C3%A1rio) , unfortunately in portuguese), that explores this very issue. When he was young, the director tried to shoot a do
191.
▲
by
alextp
16y ago
But these places arise precisely from the fact that they're not empty. Most chinatowns/koreatowns are in places where, back when the immigration wave was happenning, there was cheap housing that was close enough to local commerce. As soon
192.
▲
by
alextp
16y ago
This is not a trivial question when you touch on mining-related issues. One can only become that rich by an artificial scarcity that is created when one person/company "owns" a part of land it bought way below value (since the true value sh
193.
▲
by
alextp
16y ago
If you feed it words and appropriate features from a programming language you should get something close to a tokenizer. Laura dietz ( http://www.mpi-inf.mpg.de/~dietz/index.html#publications ) has some work that applies similar technique
194.
▲
by
alextp
16y ago
It is still an open problem to do unsupervised or mostly-unsupervised part-of-speech tagging that performs as well as (or close to) the usual supervised models. For English in the standard domains for which there are annotated corpora this
195.
▲
by
alextp
16y ago
That's a very thorough description. It's always great to see laymen-readable expositions of research. I find it nice that a purely functional language can handle this sort of problem (state management) more cleanly than what I usually do wi
196.
▲
by
alextp
16y ago
Minimizing context-switches and exploiting caching sure helps get housework done a lot faster and with less hassle (be it doing the dishes, sweeping the floor, doing laundry, or tidying up the place). Also sort-then-process (which can be n
197.
▲
by
alextp
16y ago
I find that a lot of people seem to be more comfortable jumping from "I have a problem" to "let's make a GA" than to "let me formulate it as an optimization problem and optimize it". Which is funny, since GAs are essentially odd optimizatio
198.
▲
by
alextp
16y ago
It was screwed for me until I realized I was using different font sizes in my browser for proportional (13) and mono-spaced (12). Changing the size of the mono-spaced font to 13 solved it for me.
199.
▲
by
alextp
16y ago
A very good source of essays is the New Yorker magazine. I subscribe to the print version, but the online version is cheap, and many good essays are actually free on their website.
200.
▲
by
alextp
16y ago
Or to running over images of children playing in the streets. Also, tourists are going to be really scared when a cab driver just ignores one of those.
201.
▲
by
alextp
16y ago
And without any visible degradation in the quality of the stories or the comments, nor any scaling issues with downtime. Good job for the reddit folks :-)
202.
▲
by
alextp
16y ago
Also, every friend I have who tried to rent an apartment in São Paulo either failed or ended up with something wildly overpriced. If you could make this generic enough to use in other countries/cities it'd be great.
203.
▲
by
alextp
16y ago
You seem to be talking about smoothing: how many non-human four-legged animals and non-four-legged humans have you got to see before you can estimate that humans are not likely to have four legs? This is reasonably well-solved; and you eith
204.
▲
by
alextp
16y ago
A popular nlp blogger posted ( http://nlpers.blogspot.com/2010/08/readers-kill-blogs.html ) about a related problem with comments (ie, that they're getting more and more distributed).
205.
▲
by
alextp
16y ago
This is very true. I sometimes use the fact that I'm a phd student as a cover to study other, completely unrelated things (and get paid for it). It's weird to see institutional validation bias into play.
206.
▲
by
alextp
16y ago
Actually, as far as emacs things goes, org mode is one of the few that works out of the box. My 500-line .emacs has no configurations in it whatsoever for org mode.
207.
▲
by
alextp
16y ago
Org-mode is great. I'm using it now to organize an intro to CS class at my university. I write the lecture notes, pass exercises, give grades, etc. There are so many things org mode can do I'm still frequently surprised at how useful it is,
208.
▲
by
alextp
16y ago
The main thing broken about the US academic system is not tenure, I think. It's grants, and the perpetual chasing of them, the effective cause of the good teachers having no time to teach and hardly any time do to research (if they restrain
209.
▲
by
alextp
16y ago
Appel's series of Modern Compiler Implementation in * is easy enough to ready and apply, and some come with code (IIRC).
210.
▲
by
alextp
16y ago
Catholicism has a very similar sin-and-atonement-based logic, and I've seen some non-practising catholics turn green.
More ›