NCU AI Course

Recent site activity

Chapter 15. Probabilistic Reasoning over Time

Slides-1, Slides-2.

15.1. Time and Uncertainty
       States and observations
       Stationary processes and the Markov assumption
15.2. Inference in Temporal Models (Discrete variables)
       Filtering: P(X_t | e_{1:t}) by FORWARD
       Prediction: P(X_{t+k} | e_{1:t}) by FORWARD
       Smoothing: P(X_k | e_{1:t}), 1<= k <t by FORWARD+BACKWARD
       Finding the most likely sequence: argmax {X_{1:t} | e_{1:t}) by Viterbi
15.3. Hidden Markov Models (One discrete hidden variable)
       Simplified matrix algorithms:
15.4. Kalman Filters (Continuous hidden variables)
       Updating Gaussian distributions
       A simple one-dimensional example
       The general case
       Applicability of Kalman filtering
15.5. Dynamic Bayesian Networks (arbitrary number of hidden variables)
       Constructing DBNs
       Exact inference in DBNs (Unrolling, variable elimination, etc.)
       Approximate inference in DBNs (Particle Filtering)
15.6. Speech Recognition
       Speech sounds
       Words
       Sentences
       Building a speech recognizer