vonnik

61 ms·

661.
by vonnik 10y ago
This intro is pretty good, but may be a bit high level for some readers of HN. It seems like it's written for non-technical readers. The idea is that AI, ML and DL are nested like Russian dolls, each subsequent one a subset of the othe
662.
by vonnik 10y ago
Why use it, if it locks you in?
663.
How to Train a Neural Network for Wine Tasting(medium.com)
2 points by vonnik 10y ago|0 comments
664.
by vonnik 10y ago
This is trying to be funny, but it's not that funny. For a bunch of different reasons. Unless you're an insider telling an inside joke, tech isn't funny, because there's too much to explain, and explanations kill jokes.
665.
by vonnik 10y ago
Recursive neural networks have not seen wide adoption in the deep learning community. To my limited understanding, this is because their performance depends on the quality of the parser you use, which makes them fragile. Has anyone had expe
666.
by vonnik 10y ago
This is a 2015 piece, so old news. Can we note that in the headline?
667.
by vonnik 10y ago
This is, frankly, a naive way to rank deep learning projects, because Github stars are cheap. Francois Chollet, the creator of Keras, comes out with a monthly ranking that takes other factors into account, such as forks, contributors and is
668.
by vonnik 10y ago
Can we capitalize Rosling and add (2006) in the title?
669.
by vonnik 10y ago
One correct diagnosis is simply not that significant, except for the woman concerned. The real news will be A/B testing on much larger data contrasting Watson's diagnoses with those of doctors sans Watson, and checking both agains
670.
by vonnik 10y ago
Isn't it counterproductive to turn it into fuel to be burned? It seems like that would leave us at best at no net new carbon dioxide, when we really should be sucking CO2 out and sequestering it away. How about we turn it into bricks o
671.
by vonnik 10y ago
Yes, but my precise point is that startups lure talent and iterate on product better than most large corporations, with enclaves and exceptions in a few large-ish companies, and virtually none any where else. $1B is the monetary expression
672.
by vonnik 10y ago
I agree that there's a triumph of politics over economics on the water issue, but I'm not sure that it's in the direction we think: Israel blames Palestinians for West Bank water shortage http://www.jpost.com/
673.
by vonnik 10y ago
You simply don't know what you're talking about. Effortlessly , first of all, is false. Secondly, while the hiring of thousands of software engineers over the last few years was trumpeted, that doesn't mean GM knows how to m
674.
by vonnik 10y ago
Attracting the talent to build a usable product is one of the most important steps in creating any business, or line of business in the case of a large company. That talent is scarce, and tends to prefer certain companies over others. I spe
675.
by vonnik 10y ago
This is a piece of amateur speculation making huge assumptions about advances in technology, and about Uber's ability to adapt to those advances. One crucial thing to remember, when considering the threat Uber faces from traditional ca
676.
by vonnik 10y ago
In that case, macro-economics is simply disqualified from being scientific. It's almost impossible to repeat large-scale events, controlling for all variables. Have to say I'm not particularly impressed with the quality of Nautilu
677.
by vonnik 10y ago
Riffing on project no. 1, here's our post on "How to Control AI": http://deeplearning4j.org/skynet
678.
by vonnik 10y ago
Market prices are determined by the intersection of supply and demand, so it is not just the buyers who choose. Google has some say over the supply. They may increase or decrease the supply based on behavior they sense more broadly in searc
679.
by vonnik 10y ago
In other words, Google profits surge because it is using TensorFlow to accurately predict how to price ads to maximize revenue.
680.
by vonnik 10y ago
By that measure, startups are powered by AI and large corps are not. By which I mean, so many variables go into successful product development. Google arguably has the strongest AI in the world, but it has been remarkably bad at introducing
681.
by vonnik 10y ago
I'm acutely curious how we can attack the problem of detecting undisclosed AI breakthroughs, particularly those made by organizations uninterested in ever revealing them. Commercially, yes, many organizations want to boast about thei
682.
by vonnik 10y ago
how is this different than TSNE? https://lvdmaaten.github.io/tsne/ anyone looking for an explanation of word2vec may find this helpful: http://deeplearning4j.org/word2vec
683.
by vonnik 10y ago
That's a false analogy. What I'm saying is not like "Numpy is faster than Python"; it's like "Numpy is faster than some scientific computing that the creators of a Python framework hand-rolled separately."
684.
by vonnik 10y ago
Matthew Dicks is a great story teller with some good tips: http://www.matthewdicks.com/new-page-5/ https://themoth.org/storytellers/matthew-dicks
685.
by vonnik 10y ago
For many readers of HN, this article is obviously true. But so obvious that you might wonder if it's newsworthy, a rehashing of what we hear so often: * AI will take jobs > a prototype inventory checker developed by Bossa Nova Robot
686.
by vonnik 10y ago
But cuDNN itself is faster than those frameworks, so they are doing something to slow themselves down, and my statement stands.
687.
by vonnik 10y ago
No, it's not. Why does TensorFlow have separate benchmarks for Soumith's convnet comparison that show it to be significantly slower? https://github.com/soumith/convnet-benchmarks
688.
by vonnik 10y ago
Do we need two "Intel"s in the post title?
689.
by vonnik 10y ago
Upvoted. Was going to make the same point.
690.
by vonnik 10y ago
The headline is missing the word "Did" after "Accidentally", both on HN and on the post itself.

More ›