varelse

93 ms·

571.
by varelse 11y ago
Quality over quantity: sounds like a winning strategy to me.
572.
by varelse 11y ago
I can understand Amit Singhal's opposition to replacing hand-coded features with machine learning models. He's right that ML models have bizarre failure cases across large sample sizes, but he's apparently career-endingly wr
573.
by varelse 11y ago
In that case, one might as well boot up the President Dwayne Elizondo Mountain Dew Herbert Camacho Bot...
574.
by varelse 11y ago
The scuttlebutt I heard on him is that he opposed using machine learning and artificial intelligence techniques for search, specifically deep neural networks. I can understand his opposition, Google search results have been monotonically de
575.
by varelse 11y ago
This approach works IMO only when all of the terms have sufficient representation in the dataset. I have noticed that Google seems to have recently incorporated something like this as one of its scoring factors. That's fine if you&#
576.
by varelse 11y ago
My takeaway is something I've observed and commented about for some time: understanding how any important/relevant technology works gives you an unbeatable edge over most of your peers. Andrew Ng makes a similar comment about 2&#
577.
by varelse 11y ago
>Really? You don't think society places great emphasis on milestone achievements like marriage, reproduction, employment success, sporting prowess, etc? Only you know what will make you truly happy - but everyone will be very keen t
578.
by varelse 11y ago
I found that there are a lot nice people in NYC that have to put up with a minority of absolute jerks that give the place a undeserved bad rep. This part is not much different from SF IMO.
579.
by varelse 11y ago
Dry-farmed tomatoes (of any variety) are my personal favorite. http://www.thekitchn.com/what-are-dryfarmed-tomatoes-126811
580.
by varelse 11y ago
IMO (speaking from my recent disastrous stay in NYC): Pot.Kettle.Dystopia...
581.
by varelse 11y ago
What do you like about NYC? I spent a couple weeks there recently, to which I was really looking forward since I had never been there, and after about a week I loathed the noise, the garbage everywhere, and the lines at all the stores. Sh
582.
by varelse 11y ago
If you hate current day SF, you'll looooovvve NYC...
583.
by varelse 11y ago
That's a future I think we could have had, but IMO it doesn't look like it's actually coming anytime soon thanks to market fragmentation. And even if it did, in my experience, working on a machine with remote storage and comp
584.
by varelse 11y ago
That's good news. The requirement to purchase the Intel C compiler and tools like MKL before programming their processors efficiently is IMO one of the chief reasons why NVIDIA is kicking them to the curb here with 6.6 TFLOP $1,000 co
585.
by varelse 11y ago
Except that it appears to require MKL or ACML... Sigh... ./configure Defaulting to --with-buildtype=release Cannot find a CPU math library. Please specify --with-acml or --with-mkl with a path.
586.
by varelse 11y ago
1-bit SGD and insanely high minibatch sizes (8192) it would appear: which drastically reduces communication costs, making data-parallel computation scale. If so, while very cool, that's not a general solution. Scaling batch sizes of 25
587.
by varelse 11y ago
Yes, they own it, in exchange for protecting you from the 90% chance of failure (perhaps you should save your best ideas for when you've acquired enough resources to build it on your own? I don't, ideas are cheap, it's the e
588.
by varelse 11y ago
I've been ahead of the curve a couple times and I've crashed and burned because of it only to see someone else do essentially the same thing 5 or so years later and knock it out of the park. That's bad luck and bad timing IM
589.
by varelse 11y ago
Another bogus assertion: Big companies like Amazon, Google, Microsoft, Facebook, Apple etc. are constantly iterating on innovation. Most of that innovation never sees the light of day just like most startups are destined for failure. IMO t
590.
by varelse 11y ago
Utterly false dichotomy. You can innovate from a safe job(tm) as well (not that any job is safe in an age of at-will employment). The difference is that in exchange for a lower payout for success, one gets a much higher payout for failu
591.
by varelse 11y ago
I mostly agree and I find that most of the arguments put forth in favor of UBI seem relatively narcissistic in nature (Your startup fails and you end up on the street? Yeah, not my problem actually, nor would it be my problem if you spent
592.
by varelse 11y ago
Because then the right lobbyists contact either FoxNews or MSNBC (depending on who's apparently to blame) and ignite a two-minute hate about how horrible this is by trotting out a few tragedies and the sheeple get their outrage on, no?
593.
by varelse 11y ago
Except in places like SF where the victims themselves band together to maintain low inventory so they can justify their rent and eviction controls. And all because a minority of long-term residents live at below market rates and raise the h
594.
by varelse 11y ago
Drosophila is a real noisy living system, and as a biologist myself, and unlike the data scientists with whom I work, none of us would consider the fruit fly to be an approximation of a human being. Further, the biologists and chemists I kn
595.
by varelse 11y ago
I would almost state "Clean datasets (like MNIST) considered harmful" because without exception the data scientists I meet who work with them just shovel real data blindly into ML code and kvetch when they don't beat simpler
596.
by varelse 11y ago
In my experience, the STEM people are eager to learn and I've mentored several. OTOH the people that get my goat are the data scientists who only know single-threaded python code and rely on hadoop for parallel execution who insist on
597.
by varelse 11y ago
Are you kidding me? Exploiting the native warp-width is a 50-100% performance improvement when appropriate. Not querying the warp-width OTOH is amateur hour coding.
598.
by varelse 11y ago
And the OpenCL incarnations all have wretched performance relative to CUDA. Why doesn't someone do something about that?
599.
by varelse 11y ago
I got the crap kicked out of me in music class when I was 12 for playing "National Brotherhood Week" as an example of a song I liked...
600.
by varelse 11y ago
Did I say anything about keeping or even opening the packages? If the drone is using some sort of convnet or HOG to detect the landing pad, I can only imagine hilarity would ensue if its visual field were saturated with multiple copies of

More ›