When using the predict_(MatrixDouble& obs) method of the continuous hmm classifier if the downsample factor is smaller than the number of rows of the given matrix.
This is not really a bug but I expected the matrix predict function to behave in the same way as the vector predict function (that uses a circular buffer to hold real time data). Given a bunch of realtime data one may want to process it as a whole and just get the last class predicted, this could be done by setting the circular buffer directly but being a protected attribute that is not possible.
When using the predict_(MatrixDouble& obs) method of the continuous hmm classifier if the downsample factor is smaller than the number of rows of the given matrix. This is not really a bug but I expected the matrix predict function to behave in the same way as the vector predict function (that uses a circular buffer to hold real time data). Given a bunch of realtime data one may want to process it as a whole and just get the last class predicted, this could be done by setting the circular buffer directly but being a protected attribute that is not possible.