praccu

8 ms·

31.
by praccu 10y ago
I think there may be a distinction: Walmart could fill a hole in the ground with $3 billion dollars and burn it, and not care. [A bit of an exaggeration, as that's half their annual profit] AOL acquired Time Warner for $182 billion.
32.
by praccu 10y ago
https://www.ldc.upenn.edu/ http://kingline.speechocean.com/ http://deeplearning.net/datasets/ http://www.elra.info/en/
33.
by praccu 10y ago
old mathematicians
34.
by praccu 10y ago
http://www.fatml.org/
35.
by praccu 10y ago
Failure data can be useful in a more fine grained way, too: http://www.nature.com/news/computer-gleans-chemical-insight-...
36.
by praccu 10y ago
Thorsten Joachims gave a talk at Amazon Machine Learning Conference 2015, about doing specifically that. That may be what someone was talking about. I've been trying to find the paper related to the work, but am struggling to find it.
37.
by praccu 10y ago
Fascinating. The qualitative comparison suggests that the item2vec may produce _more_ homogenous / boring results, which is kinda unfortunate; the interesting question in recommendations is how to find "aspirational" recommen
38.
by praccu 10y ago
http://www.wired.com/2013/11/minecraft-book/
39.
by praccu 10y ago
Of course you include stock? That's part of the cost. I know four people that just started fresh out of no-name schools for $120 including stock.
40.
by praccu 10y ago
I know four people just hired making $120k, not from name-brand schools?
41.
by praccu 10y ago
Uh, at Amazon, junior devs go for around $120 if they don't negotiate well. Interns, last I checked, were close to $80. If you fold in the TCO (benefits, computer / chair / licenses, etc.) for a senior engineer, the $350 isn&
42.
by praccu 10y ago
This is a totally shameless plug, but here's some a PR thing about some work in that direction: http://www.businesswire.com/news/home/20160216005731/en/Voca...
43.
by praccu 10y ago
DNNs have lead to massive improvements in speech, specifically in the acoustic model. RNN based language models interpolated with modified kn smoothed ngrams (often plus some other cool stuff) are also the state of the art, although they&#
44.
by praccu 10y ago
cobaltspeech.Com -- in case you can't get what you want from a default Kaldi install.
45.
Why scientists should learn to fail, fast and often(statnews.com)
3 points by praccu 10y ago|0 comments
46.
Lab Failures Turn to Gold in Search for New Materials(scientificamerican.com)
5 points by praccu 10y ago|0 comments
47.
by praccu 10y ago
/wave Lemme know if you've got questions.
48.
by praccu 11y ago
> 'There are a lot of problems in speech besides transcription. ' I'm trying to be vague because I don't want to violate NDAs and it's a pain to think up examples I'm not currently working on, but there are
49.
by praccu 11y ago
There are a lot of problems in speech besides transcription. I should have been clearer about what I meant. Even for people that do want transcription, though, often you can handily beat the accuracy of off-the-shelf services by tuning the
50.
by praccu 11y ago
Some state-of-the-art industrial speech recognition [0] is transitioning from HMM-DNN systems to "CTC" (connectionist temporal classification), i.e., basically LSTMs. Kaldi is working on "nnet3" which moves to CTC, as we
51.
by praccu 11y ago
Yes. Language learning, for example. Speech recognition is being used for a lot of IoT stuff, home automation, and toys and the like. Lots of other really interesting niche applications. We do research consulting in speech & language, a
52.
by praccu 11y ago
Huh? The linked paper is 4 years old. DNNs have been dominant in speech since 2012. No one uses GMM systems anymore. Baidu's approach isn't even the best (IBM's system tends to beat theirs on accuracy, and google tends not to
53.
by praccu 11y ago
Kaldi is the best, in my opinion. It has a few open acoustic models. Unfortunately, speech is just hard; acoustic models are onlying one part of these systems.
54.
by praccu 11y ago
aah. Okay. Disclaimer, [b] this isn't my area [/b] and I'm going to miss some folks. I apologize to all of the important and interesting work that I'm missing. Also, I'm really linking to research groups, so take th
55.
by praccu 11y ago
Did you try the chipmunk voice?
56.
by praccu 11y ago
Ivona, acquired by Amazon, is a hugely notable example. They're great folks, and do great work. Alexa'a voice was done by them, but an early customer of theirs was the Polish public transit system. A lot of really great work is
57.
by praccu 11y ago
http://www.cobaltspeech.com ?
58.
by praccu 12y ago
It's interesting that graphics and objects are so connected, in your posts. I'll say that in general, EVERYTHING in Python is objects, and they're an extremely general programming paradigm. It sounds like Haskell might be a l
59.
by praccu 12y ago
Jumping (back) onto the Kendall food comment train: Other places I haven't seen mentioned: -- Squeaky Beaker -- Boca Grande -- Take-out from Similans Also, what happened to the MIT food court and when is it coming back? I want terrible
60.
by praccu 12y ago
Momogoose for life. There are a bunch of food trucks behind the T stop. Also, check out Aceituna.

More ›