HMM

Question 1

HMM

Answer 1

• Hidden Markov Model
• model for data arranged over regular time intervals
• temporal dependence of observable outcome, handled by temporal evolution of ‘hidden’ quantity

Question 2

Transition Matrix

Answer 2

• probabilities p_ij for all combinations of i and j
• each row - current state
• each col - future states
• infer how likely transitions between states are - based on prior information

Question 3

HMM - Markov Property

Answer 3

probability of being in state i at time t+1 depend only on current state

Question 4

Forward Algorithm

Answer 4

• analytical solution to integrating out the latent state sequence
• compute marginal likelihoods efficiently
• recursively calculating the probability of reaching each state at each time point, taking into account the sequence up until that point