hardmaru

72 ms·

1. 781.▲
   Solar-powered plane completes journey across Pacific Ocean(phys.org)

   1 points by hardmaru 11y ago|0 comments
2. 782.▲
   Five Nights in Pyongyang: Inside the Secret Brothel of North Korea’s Elite(eastasiatribune.com)

   3 points by hardmaru 11y ago|2 comments
3. 783.▲
   The Frog of CIFAR 10(blog.otoro.net)

   1 points by hardmaru 11y ago|0 comments
4. 784.▲
   by hardmaru 11y ago

   It's very interesting you bring this up, Canjie (倉頡) was the first Chinese input method I learned, and I didn't know it was designed to run within disk+memory limitations. To this day, I think it's still the best input metho
5. 785.▲
   by hardmaru 11y ago

   A blog post on how this algorithm works: http://blog.otoro.net/2015/12/28/recurrent-net-dreams-up-fak...
6. 786.▲
   Show HN: Neural Net Writes Fake Kanji in Your Web Browser, Stroke-By-Stroke(otoro.net)

   16 points by hardmaru 11y ago|3 comments
7. 787.▲
   by hardmaru 11y ago

   Hi Can you clarify the difference between autodiff and symbolic diff? Thanks
8. 788.▲
   by hardmaru 11y ago

   Maybe this can be another machine learning problem.
9. 789.▲
   by hardmaru 11y ago

   The problem with this particular algorithm is that I needed to have a lot of data to make it sort of work. I'm not sure how your AI was trained to play that game, sounds like a difficult problem-
10. 790.▲
    by hardmaru 11y ago

    Is that why the shape of the fake pen for that Bimoji Kanji Writing NDS video game shaped like a brush?
11. 791.▲
    by hardmaru 11y ago

    We were actually sort of discussing this a short while back https://twitter.com/hardmaru/status/676067023655337984
12. 792.▲
    by hardmaru 11y ago

    Cool, I wonder how difficult it would be to extract all of the stroke data. LINE would probably not put the data on github..
13. 793.▲
    by hardmaru 11y ago

    I reproduced an English handwriting generation experiment in the previou post. http://blog.otoro.net/2015/12/12/handwriting-generation-demo...
14. 794.▲
    by hardmaru 11y ago

    maybe I went too far with the cultural jokes compared to typical papers written on machine learning- I meant no harm, hopefully I didn't offend too many! I love living in Japan, and everything about Japanese culture (even Japanese comp
15. 795.▲
    by hardmaru 11y ago

    my grandfather would turn over in his grave if he saw my Chinese handwriting too :)
16. 796.▲
    by hardmaru 11y ago

    Hi Kuschku Can you let me know the resolution of your device? I'm trying to play around with the CSS for blog.otoro.net, and trying to find a good balance for different resolutions. Thanks
17. 797.▲
    by hardmaru 11y ago

    Yeah reinforcement learning and policy gradients can possibly attack the same problem. Although for a simple task like this demo, I stand by simple conventional neuroevolution as they would work best and are easy to train as the nets are qu
18. 798.▲
    by hardmaru 11y ago

    Yeah, I can make them have 16 but it slows things down esp on mobile phones.
19. 799.▲
    by hardmaru 11y ago

    I guess (1) or (2) would depend on the culture that gets evolved :)
20. 800.▲
    by hardmaru 11y ago

    The network would be too small to develop this complex behavior. I had to balance between having it run well in js and the computation complexity of each creature.
21. 801.▲
    by hardmaru 11y ago

    I would love to see the youtube for that too :) Are you sure you were not also stoned
22. 802.▲
    by hardmaru 11y ago

    Thanks for the feedback, I'll need to change around the CSS later.
23. 803.▲
    Show HN: Neurogram – Creating art by evolving neural networks in Javascript(blog.otoro.net)

    7 points by hardmaru 11y ago|0 comments
24. 804.▲
    Show HN: Critters with neural net brains coevolve to survive killer planks in JS(otoro.net)

    65 points by hardmaru 11y ago|5 comments
25. 805.▲
    by hardmaru 11y ago

    Hi This post outlines a version that doesn't start pre-trained http://blog.otoro.net/2015/05/13/neural-slime-volleyball-evo...
26. 806.▲
    by hardmaru 11y ago

    cool! Let's talk further offline later- I would be interested to know what you end up doing with it, and pick your (biological) brain for ideas. Ping me on Twitter message later-
27. 807.▲
    by hardmaru 11y ago

    Thanks for the references So you think basically recording human actions (possibly in some feature extracted way) in part of the 'replay memory' for DQN would work well? I have also been reading about deepmind's recent survey
28. 808.▲
    by hardmaru 11y ago

    Posted a small write up here with regards to the evolution setup used: http://blog.otoro.net/2015/05/13/neural-slime-volleyball-evo...
29. 809.▲
    by hardmaru 11y ago

    The creatures evolved much faster! If you run the training demo at http://otoro.net/planks/training.html I noticed that after 50-100 generations they develop insect like characteristics. But for the slime agents, they
30. 810.▲
    by hardmaru 11y ago

    I should have done a better job for the ball dynamics Perhaps the easy version is more suitable for the kids http://otoro.net/slimevolley/easy.html Thanks for the feedback it is very helpful.