On Wed, Mar 17, 2010 at 10:25 PM, Jason Grout <jason-s...@creativetrax.com> wrote: > On 03/17/2010 10:38 PM, William Stein wrote: >> >> Hi, >> >> I'm curious if anybody reading this uses the Hidden Markov Models >> (HMM) code in Sage for anything. If so, send me an email, since I'm >> working on 100% replacing it by some brand new better quality code. >> >> In case you don't know, HMM's are a tool in statistics and machine >> learning, which play a big role in speech recognition and analysis of >> structure in DNA sequences (bioinformatics), among other things (such >> as modeling certain sequences that appear in analytic number theory). > > > I used them in my modeling class this last semester (and then a student used > them for their final class project, and another student did a project > involving HMM on CUDA GPUs for their senior project, though not using Sage). > > I saw your patch. I'm curious why you are doing this. It seems that your > attitude when you first interfaced with ghmm was that you'd rather spend a > few days writing interface code to an existing small fast library, rather > than recoding the algorithms by hand. Your patch now seems like a total > turn-around in attitude.
I was wrong. It is stupid to spend a few days writing interface code to an existing library, when one can completely implement the algorithms in that library better from scratch in a few days. The quality of the GHMM library is "questionable", and I didn't realize that the difficulty of coding HMM-related algorithms is much, much lower than the difficulty of coding most number theory (and exact linear algebra) algorithms, which is what I usually work on. The main problem was probably that I didn't sufficiently conceptually understand the algorithms well enough before. I'm not going to elaborate on the quality issues with GHMM, since I don't see any point in offending the authors of that library. -- William -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
