gnipgnip

18 ms·

1. 121.▲
   by gnipgnip 10y ago

   https://en.wikipedia.org/wiki/Graphical_model
2. 122.▲
   by gnipgnip 10y ago

   - It appears there are indeed things like dataflow analysis/ SMT that can achieve better performance over current techniques. - Inference on loops seems to be something that can be handled as well (dynamically ?).
3. 123.▲
   by gnipgnip 10y ago

   Yes, this is precisely what is called probabilistic inference. I was trying to find out what more it can do other than parsing a DSL into a graphical model.
4. 124.▲
   by gnipgnip 10y ago

   Can someone in the know please comment about how this is useful ? I mean yes this is neat, but inference engines already work declaratively with factor-graphs and potentials. Are there techniques exploiting algebra/logic that helps mak
5. 125.▲
   by gnipgnip 10y ago

   Sweet AI turning a profit ? I must've been a dove-eyed kid when that last happened. Won't they have to fly back over the Atlantic from SFO, though ? Also, if the AI boffins are listening: - "एअर् " -> "अय्र्&quot
6. 126.▲
   by gnipgnip 10y ago

   Somehow, everyone thought that putting people in a straight-jacket for 30-35 years is a great way for obtaining "creative individuals".
7. 127.▲
   by gnipgnip 10y ago

   The pressure head will likely fall off very rapidly with the number of gates.
8. 128.▲
   by gnipgnip 10y ago

   Terrible title. I came here hoping for mystical stuff like, http://www.personal.psu.edu/~tms9/Therapy_Waters.html Also. Look everyone! Programmable electrons!
9. 129.▲
   by gnipgnip 10y ago

   In the US, passenger rail has always been neglected. This is not uniformly the case around the world.
10. 130.▲
    by gnipgnip 10y ago

    I'm surprised they couldn't find a buyer over in India (and still keep the engineers on board).
11. 131.▲
    by gnipgnip 10y ago

    Here's a very useful blog that deals exactly with this topic, http://www.computervisionblog.com/2016/01/why-slam-matters-f... I'd advise others on this thread, who tend to dismiss the difficulty of visio
12. 132.▲
    by gnipgnip 10y ago

    Here's the thing. Even our most advanced ML models are quite simple mathematically.
13. 133.▲
    by gnipgnip 10y ago

    Could you share with us what the state of art in SLAM is ? Last I looked, the backend is essentially the same old Kalman-smoother, but the frontend is generally used with hand-tuned features (stuff from the 90s - 00s). Big Data, incl. all t
14. 134.▲
    by gnipgnip 10y ago

    Part of the problem I think is that the image is not fully used; I mean, generally these systems consist of a black-box routine that extracts interest points and then passes it onto to a SLAM routine, which inturn keeps an estimate of the c
15. 135.▲
    by gnipgnip 10y ago

    Hmm. Is Turi going to be separate from Apple AI division ?
16. 136.▲
    by gnipgnip 10y ago

    Myanmar, DPRK, Vietnam etc. were less "socialistic" than India ? Words, words, words, how they trap us all.
17. 137.▲
    by gnipgnip 10y ago

    India is not really open; sure it's a "democracy", but it's really a de-facto feudal state, often with very little private property rights, and a bureaucracy that would put the Vogons to shame. It's not clear if con
18. 138.▲
    by gnipgnip 10y ago

    See, but the politicians didn't create the system; they essentially compete to be the top Vogon. I'd discount some of the paranoia around the FRB, but the modern monetary system appears to be a very very clever way of herding a po
19. 139.▲
    by gnipgnip 10y ago

    Not to mention many results in optimization/computer-science came out of Russia.
20. 140.▲
    by gnipgnip 10y ago

    Awesome work China!
21. 141.▲
    by gnipgnip 10y ago

    The past few years have been a re-education for me on Language, Lies, Money, Propaganda and much else. I remain in awe of the state apparatus, and the cynical brilliance required to run it all (assuming there is such a central clique).
22. 142.▲
    by gnipgnip 10y ago

    I don't think you understand how money works.
23. 143.▲
    by gnipgnip 10y ago

    I wish papers were "running", as in a wiki page with developments being incrementally added. It'd certainly help reduce the amount of redundant reading, and much greater coherency.
24. 144.▲
    by gnipgnip 10y ago

    Interesting. Alt-news folk backing Trump will have a heart attack upon seeing the steering committee head of Bildeburg back him.
25. 145.▲
    by gnipgnip 10y ago

    I think it's producing a lot of technology; knowledge, I'm not so sure.
26. 146.▲
    by gnipgnip 10y ago

    A Servo ? Some reason for why a relay wasn't used ?
27. 147.▲
    by gnipgnip 10y ago

    :) https://youtu.be/5YBwDNfOaxU?t=16s
28. 148.▲
    by gnipgnip 10y ago

    I'm not familiar with medicine; was this not known in the field ? Surely a student from Stanford, who I hear was also doing research at NUS, should have had access to enough advisers who could've averred that to her (before the co
29. 149.▲
    by gnipgnip 10y ago

    Causality is a tricky thing (see Pearlmutter et.al); often people jump on the most socially-salacious (Latin salire ~ "leaping") explanation, after the fact, probably because it is at that moment, the best available instrument for
30. 150.▲
    by gnipgnip 10y ago

    Erm, you can't calibrate out Boltzmann noise.