currymj

81 ms·

901.
by currymj 8y ago
“It takes a big man to admit his data is small.”
902.
by currymj 8y ago
they want to get well-paid, high prestige jobs using machine learning. whereas doing the exact same task with a regression in Excel turns it into low-prestige clerical work. i think they’re acting pretty rationally once you consider their o
903.
by currymj 9y ago
Massachusetts will give priority in getting licenses to marijuana retailers that employ people with prior drug convictions. additionally, if the owners (or immediate family members) themselves have drug convictions there is assistance wit
904.
by currymj 9y ago
the gambling industry pioneered a lot of the techniques now used for web sites, far earlier. “time on device” is something they coined and started optimizing for decades ago.
905.
by currymj 9y ago
i'm not sure about plumbers, but i think CS graduates do much worse in the UK, even adjusting for cost of living and currency and so on. of course it's hard to compare statistics, but 40k pounds a year would be somewhere around no
906.
by currymj 9y ago
it is an extremely fun game! especially for all the oldheads around here, if you ever played Zork or Hitchhiker's Guide or whatever, give it a shot. it's the same style, just as funny, with helpful features to make it far less fru
907.
by currymj 9y ago
i don’t know that it has to be competitive in speed right away as long as it works and is easy to install. i’d be thrilled to have the option of doing deep learning on an AMD card even if it ran at 1/3 the speed on a comparable (but pr
908.
by currymj 9y ago
i tend to agree that baseball shouldn't be tweaked. although I think if there's a way to incentivize small ball somehow without doing anything ridiculous, that would probably be good for the sport. the three true outcomes are dull
909.
by currymj 9y ago
They are working on it pretty actively [1], you can see a progress chart on their website. I think some of the PyTorch devs have said they'd welcome AMD compatibility. If you go to the ROCmSoftwarePlatform github page you can see some
910.
by currymj 9y ago
i also raised my eyebrows at this. i think "workstation" might have been a better choice of word than "personal computer". but people, especially in the sciences, do actually buy desktop computers with these sorts of spe
911.
by currymj 9y ago
part of the reason may just be that there's only one season a year, and you can't reasonably change the rules in the middle. i assume game devs are perfectly willing to push out balance adjustments more than once a year. likewise
912.
by currymj 9y ago
https://www.youtube.com/watch?v=jxl6IvDvGHU That's a video where they lay out how they want it to work, if you're interested and haven't seen it.
913.
by currymj 9y ago
it's true of many doctors as well. the difference is both doctors and bankers do it for a few years at the beginning of their careers, and in exchange are promised a path to a real reward -- they get promoted, and their pay increases i
914.
by currymj 9y ago
one pretty clearly inspired by Julia's approach, from what I can understand.
915.
by currymj 9y ago
it’s absolutely true, but this kind of statement is why people get so enraged at New Yorkers talking about local food. without context it sounds like you are saying putting butter on a roll is a New York innovation.
916.
by currymj 9y ago
it's kind of a subjective judgment. it's easier to define bad ergonomics, I think -- it's where the way part of the language is designed requires you to write very contorted, strange code to accomplish a task that ought to be
917.
by currymj 9y ago
i think a factor is that a lot of New York small businesses, food trucks, etc. buy from the same local suppliers, and because New York is such a big market, it's actually viable for them to only serve one city. The most visible example
918.
by currymj 9y ago
this is probably the most infuriating thing about academic CS, is condescension from people in every other field about this fact.
919.
by currymj 9y ago
this is the right way to go about things if you have certain goals, for sure. sometimes you need to replicate exactly the same training method, on exactly the same data — for instance if you want to use it as a baseline on a known dataset.
920.
by currymj 9y ago
you’d be surprised. in the past at least there’s been far less oversight of real estate transactions than other types.
921.
by currymj 9y ago
yeah, the idea here is that it might be sufficient, much like cryogenics. their marketing is probably irresponsible, depending on your perspective, but then again you're dead anyway so any shot at a future may be worth it. unlike cryog
922.
by currymj 9y ago
David Silver (now of DeepMind, leading their reinforcement learning group which has had so many big high profile successes) has video lectures for his introduction to reinforcement learning. It's very thorough and uses a good, free tex
923.
by currymj 9y ago
> Could you be more concrete about how you would like other fields to contribute? In the world of fair machine learning, "What are the right criteria for fairness under [set of circumstances]?" is usually not easily answerable,
924.
by currymj 9y ago
the first problem is one where CS researchers would likely benefit from more contact with outsiders, although I agree it's a pretty lively area of research already, at least for the shorter-term issues. the second problem is one where
925.
by currymj 9y ago
i think this is a fair concern, especially when you consider the typically abysmal quality of science journalism about machine learning. but I also don't think someone trained in CS, who has a deep understanding of machine learning sys
926.
by currymj 9y ago
if a sociologist wanted to study the social impact of transit infrastructure, that would be perfectly reasonable, and we wouldn't expect them to know much going in about signaling systems, switches, or different types of rolling stock.
927.
by currymj 9y ago
people would assume you paid for them, which minor public figures like journalists, business owners, and politicians are often mocked for doing.
928.
by currymj 9y ago
Democrats do not all want single payer healthcare. Only the left wing of the party really wants it, although this is slowly changing. Center and center-right Democrats are pretty happy with the ACA as it was intended, although often not as
929.
by currymj 9y ago
i've seen a number of math-but-sort-of-CS textbooks putting the dependency graph for chapters in the introduction, so you know what you can get away with skipping and what must be done first. i agree about geometric intuition, as long
930.
by currymj 9y ago
friendly conversation, sure, but there's enough people who refuse to take a hint that their romantic overture is unwanted. they might even retaliate with really aggressive, angry behavior. this is rare, but i think it has poisoned the

More ›