3
$\begingroup$

How could I use a 16x16 image as an input in a HMM? And at the same time how would I train it? Can I use backpropagation?

$\endgroup$
5
$\begingroup$

You wouldn't, normally. A HMM is used to model sequences of observations, and it would not make sense to use it for image recognition. Unless they are sequential, such as strokes in handwriting.

HMMs are typically used in fields such as speech recognition and part-of-speech tagging. Here you observe a sequence of events that you want to fit to a model in order to classify the individual observations.

For training a HMM you would use something like the Baum-Welch Algorithm; for finding the most likely sequence (ie the recognition process) the Viterbi Algorithm is used.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Anyway hypothetically would it be possible? $\endgroup$ – david david Apr 7 at 12:00
  • $\begingroup$ @daviddavid No, it's the wrong tool. $\endgroup$ – Oliver Mason Apr 7 at 12:14
  • $\begingroup$ What if we thought about pixels as sequential data? $\endgroup$ – david david Jun 3 at 11:45
  • $\begingroup$ @daviddavid That wouldn't make sense, as they are presumably unrelated, and not a meaningful sequence. Ultimately that depends on the type of image; when pixels are determined by the pixels around them, then it might just about be possible. $\endgroup$ – Oliver Mason Jun 3 at 11:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.