HMM VariantsΒΆ

Sequentia provides two types of HMMs that can be used with a HMM Classifier.

  • Gaussian Mixture HMM:
    For modelling univariate or multivariate numerical sequences with real-valued observations.
    e.g. MFCCs, sEMG/ECG/EEG signals, position/rotation signals, word embeddings.
  • Categorical HMM:
    For modelling univariate categorical sequences with observations from a finite set of symbols.
    e.g. DNA sequences, part-of-speech tags.