Hidden Markov Models (HMM): Theory

This lecture explores the concept of Hidden Markov Models (HMM) as a method for modeling time series data. Starting with the discussion on the time dependency of data and the challenges of unstructured trajectories, the instructor explains how to model data using HMMs. The lecture covers the parameters of HMMs, learning HMMs, computing the likelihood of an HMM, and using HMMs for prediction. Additionally, the Viterbi Algorithm for decoding in HMMs is introduced, along with the process of choosing the number of states in an HMM. The lecture concludes with a summary highlighting the significance of HMMs in modeling time series data.