[mlpack-git] [mlpack/mlpack] Implement multidimensional discrete distribution (#810)

[Ryan Curtin]

See #809 for the discussion that led to this ticket.

Currently, the `DiscreteDistribution` class in `src/mlpack/core/dists/` only supports single-dimension discrete inputs (like `(1)`, `(2)`, `(5)`, etc.).  It would be very useful if it instead supported multi-dimensional discrete inputs (like `(1, 2, 3)`, `(3, 5, 8)`, `(1, 7, 5)`).  So the `DiscreteDistribution` class can easily be extended to support that case.  Right now, in the single dimension case, the probability of some input X is simply stored as P(X).  Now, with multi-dimensional inputs x_1, x_2, ..., x_d, the probability should be P(x_1) P(x_2) ... P(x_d) (just single-dimension probabilities multiplied).

When we close this ticket it should be possible to train multi-dimensional discrete HMMs with the `mlpack_hmm_train` program, and there should also be tests for the multi-dimensional support in `src/mlpack/tests/distribution_test.cpp`.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/810
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20161108/ac0da922/attachment.html>