Working Hybrid #1802
Working Hybrid #1802
Conversation
|
@varunagrawal let me know when the examples are in here and I should look again. |
|
PS, avoid pushing each commit to save on CI cycles :-) |
…GaussianMixtureModel)
…ovariances for later
| @@ -115,6 +118,156 @@ TEST(MixtureFactor, Dim) { | |||
| EXPECT_LONGS_EQUAL(1, mixtureFactor.dim()); | |||
| } | |||
|
|
|||
| /* ************************************************************************* */ | |||
| // Test components with differing means | |||
There was a problem hiding this comment.
I need more explanation in the comment to understand these tests. Also, since in the end they are just regression tests, it’s unclear they assert correctness of anything :-/ if these examples are supposed to be equivalent to the HBN tests above, and you are using the same values as there to test the result, that is a different story. Unclear now, though.
There was a problem hiding this comment.
I should remove these since they are beyond the scope of this PR.
There was a problem hiding this comment.
Did you end up removing them?
There was a problem hiding this comment.
Just removed them.
|
seems you have a compile issue. Could you maybe only push once a review is needed? That will eliminate some green-house gases when CI is run for entire matrix. |
|
Assuming the not 50-50 motion model test is the last bit. Let me know by email when I should look again. |
Yup, working on that. |
|
CI passed, want to merge? |
GaussianMixtureFactorto store the Gaussian normalizers as detailed in my thesis.DifferingMeansandDifferingCovarianceswhich verify the mode selection.logProbabilitiesusing a scheme detailed in my thesis.