aub3bhat

10 ms·

1. 61.▲
   by aub3bhat 9y ago

   I am developing Deep Video Analytics [1] which incorporates FaceNet [2] and MTCNN detector. 1. What makes you think that governments would openly release their recognition models? 2. Do you have any hard numbers? 3. What's the differen
2. 62.▲
   by aub3bhat 9y ago

   Meh DFW is the last person you ought to listen to for advice. [1] It's just demotivational-speak paraded around as if it were some profound wisdom, just to assuage ego of your average reader with Liberal arts degree. So that she/h
3. 63.▲
   by aub3bhat 9y ago

   I am building Deep Video Analytics which aims to become Database+Processing framework for visual data. With DVA you can load, store, decode, annotate, apply inbuilt and custom Deep learning algorithms. Further you can examine the data thro
4. 64.▲
   by aub3bhat 9y ago

   The article is typical clickbait, with little content let alone nuanced analysis. For those looking for a quick technical overview of SLAM take a look at blog post (by now head of vision at Magic Leap) http://www.computervisionbl
5. 65.▲
   by aub3bhat 9y ago

   Having suffered through unbelievably atrocious mobile news sites, I now refuse to open any news website that's NOT AMP. Those criticizing AMP have it wrong, Google could have very well just bypassed AMP for "instant publisher app&
6. 66.▲
   by aub3bhat 9y ago

   HN/BitcoinTalk/etc, suffer from extreme (pro-crypto) filter bubble. The average man/woman on the street ("society") could care less about Bitcoin/Ether. Just look at the whole Net Neutrality debate. Bitcoin&#x2
7. 67.▲
   by aub3bhat 9y ago

   It's also possible that the government may declare some bills invalid (what happened in India with larger bills). Also, government could inflate away your wealth by printing more currency. At end of the day governments contro
8. 68.▲
   by aub3bhat 9y ago

   Yet another article filled with feel-goodery truthiness that cannot be debunked.
9. 69.▲
   by aub3bhat 9y ago

   This looks like a really cool work for a beginner and I do not mean to diminish his effort in any way. But face recognition is a sensitive/politically-charged topic, I know several grad students (including me) who inspite of having rea
10. 70.▲
    by aub3bhat 9y ago

    75$ is too expensive for something that costs 0$. Also most developers in SV are paid more than physicians (internists/family-medicine) who prescribe flu-shots.
11. 71.▲
    by aub3bhat 9y ago

    >> Identify location of text in image. We have hundreds of images uploaded to our app each day. The issue is some of these images have text that we want cropped out. There you go [1] originally from [2] [1] https://github.c
12. 72.▲
    by aub3bhat 9y ago

    Thanks, great to see that TF recongizing the issue with Pipelines/queues and simplifying it. I wish they did NCHW conversion by automatically detecting GPU. I think Tensorflow is developed in a very principled manner focusing on abilit
13. 73.▲
    by aub3bhat 9y ago

    I actually strongly agree with him, In fact the progress in Deep Learning has stalled due to inefficiencies in sharing data and code between researchers. Today to replicate a study you have to spend several days downloading data, making sur
14. 74.▲
    by aub3bhat 9y ago

    This was reply to the grand-parent comment, mistakenly posted to parent comment.
15. 75.▲
    by aub3bhat 9y ago

    Centralization is an artifact of human societies (programmers & organizations) decision making process not result of existence/lack of decentralization features.
16. 76.▲
    by aub3bhat 9y ago

    The legal system does not "decides" on legality unless its forced too. Consider the case of Google Books project. Eventually the courts did rule that it constituted fair use. Or consider the situation involving Flickr and Pinteres
17. 77.▲
    by aub3bhat 9y ago

    I have made this exact argument (quoting Taleb) but HN users are not receptive to anything other than a do-nothing approach. https://news.ycombinator.com/item?id=14262011
18. 78.▲
    by aub3bhat 9y ago

    There was a great talk at USENIX on how Whatsapp reduced spam while maintaining end to end encryption https://youtu.be/LBTOKlrhKXk
19. 79.▲
    by aub3bhat 9y ago

    I think the real battle is in Judicial reforms. Honestly we are at a stage where the face recpgnition technology is just too easy to deploy. [1, 2] There are also several legitimate use cases, e.g. Amber alerts, Public right to record, etc.
20. 80.▲
    by aub3bhat 9y ago

    Its even closer to systems, optimization etc. In fact I would argue that methods like backpropagation, dropout, batch-normalization, LSTM, RNNs make it much closer to Optimization rather than Statistics.
21. 81.▲
    by aub3bhat 9y ago

    There is a good paper by Leo Breiman [1] which discusses the exact issue that you mentioned. E.g. Statisticians believe that there is a "True" model that generates the data, and the errors observed are merely noise. ML on the othe
22. 82.▲
    by aub3bhat 9y ago

    If you were to simply look at the overwhelming evidence (rather than resorting to use of vague words "absolutely" etc.) e.g. department affiliations of ML researchers (presenting at ICML/NIPS) and practitioners (at Google&#x2
23. 83.▲
    by aub3bhat 9y ago

    >> Machine Learning was a branch of Statistics Thats incorrect. """ Machine learning is the subfield of computer science that, according to Arthur Samuel in 1959, gives "computers the ability to learn without bein
24. 84.▲
    by aub3bhat 9y ago

    This debate is meaningless for several reasons: 1. The original argument is a strawman. What do they mean by "data"? Is it survey results, microarrays, "Natural" images, "Natural" language text or readings from
25. 85.▲
    by aub3bhat 9y ago

    There are so many issues with this post, let me enumerate: 1. Straw man tweet by some non-practitioner which is used to set up the straw-man argument. 2. The whole Digits example is ridiculous, statisticians "love" toy problems to
26. 86.▲
    by aub3bhat 9y ago

    This trend is also evident in conferences like ICML, CVPR and EMNLP.[1] I have a semi-popular open source ML project [2] and the repo gets more Chinese visitors/users (from posts on Weibo/WeChat etc.) than all other countries incl
27. 87.▲
    by aub3bhat 9y ago

    Their source code (DeepDive) is Open and Apache licensed. My guess the DARPA funding prevented FoundationDB type fate. https://github.com/HazyResearch/deepdive http://cs.stanford.edu/people/chrismr
28. 88.▲
    by aub3bhat 9y ago

    This is a spectacularly bad advice, which "off-the-shelves" research libraries that you mention? OpenCV, LibSVM, Weka or Matlab/Octave?? Most examples/implementations in OpenCV are outpaced by Deep Learning methods. Tens
29. 89.▲
    by aub3bhat 9y ago

    Sorry, I can't seem to edit it to remove. But will keep this in mind in future. Best,
30. 90.▲
    by aub3bhat 9y ago

    "fearmongering at a scale unseen before" is reasonable response when the argument uses words like "dystopian".