mattyo

I'm doing some pattern recognition and diving a bit into mnm, and I was wondering how people had approached this problem:

mnm puts out a list of weighted guesses as to which gesture you are most likely making. The most favored gesture is not inevitably the correst one, especially early on in the process. I've tried various brute-force approaches to picking the right guess: blocking choices for a short period after activating decoding, waiting for it to have settled on one decision for a certain period of time, etc, but those strategies stroingl limit the kinds of gestures that it will get. For example, variables like total time of gesture and number of changes of direction become hevily constrained in such circumstances. Has anyone come up with a smarter approach?

mnm puts out a list of weighted guesses as to which gesture you are most likely making.  The most favored gesture is not inevitably the correst one, especially early on in the process.  I've tried various brute-force approaches to picking the right guess: blocking choices for a short period after activating decoding, waiting for it to have settled on one decision for a certain period of time, etc, but those strategies stroingl limit the kinds of gestures that it will get. For example, variables like total time of gesture and number of changes of direction become hevily constrained in such circumstances.  Has anyone come up with a smarter approach?

when is mnm done recognizing