extremelearning

9 ms·

1. 61.▲
   by extremelearning 8y ago

   Absolutely! This is actually the original reason I began exploration into quasirandom numbers. I wanted to see if I could replace some of the stochastic process in current machine learning architectures with quasirandom processes. In princi
2. 62.▲
   by extremelearning 8y ago

   I somewhat stumbled on this during my efforts to improve some machine learning architecture I was working on, so unfortunately no formal proofs as yet. This is one reason why I went down the route of technical blog post rather than more ac
3. 63.▲
   by extremelearning 8y ago

   Author here. OMG! Someone submitted my post here and it's got to the front page!! Happy to try to answer any questions that people have. Enjoy!
4. 64.▲
   by extremelearning 8y ago

   Hi Jacob, I never got to thank you for doing the observableHQ notebook for my quasirandom sequencing. For your information, I have linked to it on many occasions which includes in my post "Unreasonable Effectiveness of Quasirandom Sequ
5. 65.▲
   by extremelearning 8y ago

   Interesting question... I just calculated that for the modified Fibonacci sequence (last method): 2.95 < d_N < 3.12. Technically d* does not exist, because as N-> infinity, d* alternates between 3.03 and 3.07 (depending on if k is
6. 66.▲
   by extremelearning 8y ago

   One of the reasons why this field is still so active is because different measures and methods often result in slightly different optimal configurations. In the case of this post, I focused on direct construction methods; and measures rela
7. 67.▲
   by extremelearning 8y ago

   The values for the tetrahedron, cube, octahedron, dodecahedron and icosahedron are:3.27, 3.27, 3.46,3.19 and 3.64, resp. These produce the largest d for N=4,8,6,12 and 20 resp. Thus, it is presumed by almost everyone that d=3.64 is the glo
8. 68.▲
   by extremelearning 8y ago

   Hi I included about a dozen references at the end of my post for overall material on this topic. I specifically chose these ones as they are very readable. My favourite paper is defintiely the first one, "A comparison of popular point
9. 69.▲
   by extremelearning 8y ago

   Thanks for the kind words. I’m glad you found it interesting and useful. I think one of the advantages of writing blog posts rather than academic articles is that they are often more readable to a wider audience as the authors can be a litt
10. 70.▲
    by extremelearning 8y ago

    Author here. So amazed to find that my blog post got featured on Hacker news. Happy to answer any questions that I can! :)
11. 71.▲
    by extremelearning 8y ago

    I also love using Voronoi diagrams as well as Delaunay diagrams (via Mathematica). My most recent exploration with them was to understand and visualize the different structures and characteristics of various low discrepancy quasirandom poin
12. 72.▲
    by extremelearning 8y ago

    As @twic and the OP discussed, the equal distribution of numbers within a defined range is naturally achieved through low discrepancy sequences (eg recurrence,Halton, Sobol, etc..) Furthermore, in ultra high-speed / low-level computing