Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
fchollet
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
32 ms
·
211.
▲
by
fchollet
12y ago
Completely agree --the fundamental problem to solve is that of motivation, and motivation will very often come from role models and mentors. The function of a mentor is not limited to providing an example of what can be learned and why it s
212.
▲
by
fchollet
12y ago
Hey, thank you for this sharp and insightful one-liner. I am keenly aware of my own ignorance, and always looking for opportunities to learn. Care to enlighten me?
213.
▲
by
fchollet
12y ago
I did not imply anything about school systems outside the western world, I used the term simply because I know close to nothing about education outside the west and therefore did not want to include it in my description. I also certainly di
214.
▲
by
fchollet
12y ago
The fundamental problem with the elementary school system in the west is that kids are given no room for self-direction whatsoever; school for them is de facto a part-time prison or "labor" camp. As a result they are trained by as
215.
▲
by
fchollet
12y ago
The best kind of HN comments: - are one line long - provide categorical judgments of absolute negativity - originate "from above", from an authority figure with a much higher intelligence and much better understanding of the topic
216.
▲
by
fchollet
12y ago
It is definitely a gender equality issue, but you have to understand that it's simply the other side of that same coin that is generally presented as "women are not equally respected in India". There is a universal correlatio
217.
▲
by
fchollet
12y ago
I can't edit anymore, so I'll reply to my own post: People are pointing out there is no known hash function that will map every word within a same distance D to the same hash. That is correct, and I was just meaning to explain the
218.
▲
by
fchollet
12y ago
Any single hash will be suboptimal; the "magic" would be to use a bunch of different hashes to partition your space into suboptimal buckets, and use all of them at once to get very good (though still approximative) query results.
219.
▲
by
fchollet
12y ago
I'm definitely not kidding. I'll point out that this is very much tractable (for instance Spotify runs a matrix factorization job every few days on a users x songs matrix, maybe 50 million by 30 million, which is several orders of
220.
▲
by
fchollet
12y ago
Here's an alternative that is more efficient at large scales (eg. if you have one order of magnitude more entries than just an English dictionary): performing Approximate Nearest Neighbor search [1] over a pre-computed Levenshtein spa
221.
▲
by
fchollet
12y ago
It is impossible to talk about overpopulation without defining some context first. - what lifestyle do you want for your population? - what processes do you have in place to achieve this lifestyle? It is often mentioned that the US lifestyl
222.
▲
by
fchollet
12y ago
They are not. The bottom 50% of people owns a fraction much smaller than 1/2 of all available wealth. What Oxfam is saying is: if you rank people by wealth on the x axis then plot their wealth of the y axis, the integral (sum, or area
223.
▲
by
fchollet
12y ago
Columbus, when he sailed West, used as reference a number of maps sourced from the Chinese empire (among others)... In particular, he might have used the 1492 Martin Behaim globe as his primary reference. [0] The Martin Behaim globe shows C
224.
▲
by
fchollet
12y ago
Yudkowsky should spend more time writing his (really good) Harry Potter fanfiction and less time messing with his (really bad) "philosophy" and "AI". That would spare humanity such cringe-worthy ridiculousness as the abo
225.
▲
Bees genetically engineered to 3D print concrete and plastic
(bldgblog.blogspot.com)
2 points
by
fchollet
12y ago
|
0 comments
226.
▲
by
fchollet
12y ago
Interesting insight, thank you! I think crowd-based approaches and machine-based approaches will always have very different strengths and weaknesses. The quality of algorithmic answers will generally (always?) be well below that of human-wr
227.
▲
by
fchollet
12y ago
On the space of all valid questions, there are many it can't answer. Judging from my logs, it's still doing pretty well on most questions. Even though people are trying hard to trick it with impossible questions. Strong AI is not
228.
▲
by
fchollet
12y ago
I am working on handling ambiguity, as well as questions that don't have one definite answer (eg. "how should I invest 10k at a 5 year term?"). Possibly through an interface that shows a spectrum of answers and their sources.
229.
▲
by
fchollet
12y ago
A discussion of the algorithm powering this can be found at: http://www.sphere-engineering.com/blog/quickanswers-io-a-new... However there are actually several types of algorithms working for different kinds of questio
230.
▲
Show HN: Knowledge engine – instant answers to any question
(quickanswers.io)
48 points
by
fchollet
12y ago
|
46 comments
231.
▲
by
fchollet
12y ago
Is it possible that this, and the release of this information, is part of a NSA strategy for deterring people from using TOR or encryption? Internet users will now be too scared to be put on some list to even look up information about TOR o
232.
▲
99/1 is the new 80/20
(rude.vc)
2 points
by
fchollet
12y ago
|
0 comments
233.
▲
by
fchollet
12y ago
Although Google has done a much better PR around its research efforts, IBM and Microsoft are both investing much more in R&D. Microsoft is the tech company with the most research investment.
234.
▲
by
fchollet
12y ago
What I'm saying is, we don't have hard evidence pro or against Karpeles, what we have is a lot of speculation, typically made by angry people who lost part of their savings (or all of it in the case of Mr. Sillysaurus) in the MtGo
235.
▲
by
fchollet
12y ago
I understand the anger coming from those who lost sizeable amounts of coins due to MtGox's mistakes, but with an undergoing police investigation, wouldn't the rational attitude here be to wait for the fog of war to clear up and th
236.
▲
by
fchollet
12y ago
Non native, 81%. I think the test might be easier for non-natives speaking european native languages, as they cumulate etymological understanding from several languages.
237.
▲
by
fchollet
12y ago
It is very difficult to get somebody to understand something when they have some amount of income depending on not understanding it. Of course site owners are always going to stand behind the ad-supported model of the web, even though it cr
238.
▲
by
fchollet
12y ago
This is a very shallow way of "escaping"; your browsing activity is definitely being linked to your social media identities via your IP(s) and browser fingerprint. For instance Google does maintain such non-cookie-based user ident
239.
▲
by
fchollet
12y ago
It is essentially a Lua equivalent to Theano. A significant difference, though, is that it makes running your NNs on GPU extremely easy. Also, it is endorsed by the main deep convnets research groups (Facebook, Google).
240.
▲
by
fchollet
12y ago
I am seeing the expected Reuters article. Mind explaining what is supposed to happen when loading this page?
More ›