A new HMM topology for shape recognition

Created on 2022-05-14T06:44:03-05:00

Return to the Index

This card pertains to a resource available on the internet.

This card can also be read via Gemini.

Ergodic models: can transition between any state to another state

Temporal models: can only transition to states further in a sequence

Circular HMM: Allow the terminal state to have transition probabilities back to the beginning states.

While a temporal model enforces a kind of "moving forward" in a sequence you are then subject to having a good beginning and end point. By connecting the end to the beginning as a mobius strip you can start at any part of the sequence and become more invariant to the start and end positions.