Hidden Markov Models Quiz

Answers and explanations appear as you take the quiz.

1. What does the 'hidden' in Hidden Markov Model refer to?
   • The states are unobservable and inferred from outputs
   • The model parameters are secret
   • The observations are encrypted
   • The transitions are random and unknowable
2. What is the Markov property assumed by an HMM?
   • All states emit the same symbol
   • The next state depends only on the current state
   • Observations are independent of states
   • Every state is equally likely
3. Which algorithm finds the single most likely sequence of hidden states?
   • The Forward algorithm
   • Baum-Welch
   • The Viterbi algorithm
   • Gradient descent
4. Which three sets of probabilities fully specify an HMM?
   • Prior, likelihood, and evidence
   • Mean, variance, and covariance
   • Input, hidden, and output weights
   • Initial state, transition, and emission probabilities
5. What does the Baum-Welch algorithm do?
   • Learns HMM parameters from data via expectation-maximization
   • Encrypts the observation sequence
   • Finds the most likely single path
   • Computes the exact transition matrix in closed form
6. Which Transformer feature can directly relate tokens at distant positions in a sequence?
   • An HMM emission table
   • Self-attention over positions in the input
   • A fixed single-state transition
   • A requirement to process only one token ever
7. What does the Forward algorithm compute for a sequence of observations?
   • Only the single most likely state path
   • The most likely parameter initialization
   • The total probability of the observations by summing over possible hidden-state paths
   • The mean of the emission probabilities
8. A hidden state emits a word or acoustic symbol. Which HMM probability describes that relation?
   • The initial-state probability
   • The transition probability
   • The discount factor
   • The emission probability