Dynamics Programming for Machine Learning- Hidden Markov Models
summary
Builds a Hidden Markov Model from first principles, implements a tiny Python Viterbi in a few lines and shows it recovers the most probable state path while explaining why memoized table filling beats naive recursion. Reports time O(S^2 T) and space O(S T), covers speech features with 25 ms windows every 10 ms, DNA exon/intron triplet structure, and a KLT image feature example, and outlines training via Baum Welch expectation maximization and Viterbi dynamic programming.
AI-written summary · about
chapters (15)
- 0:00Hidden Markov models intro
- 1:15Speaker background
- 2:18Speech recognition example
- 3:01HMM structure and properties
- 6:13Viterbi algorithm overview
- 9:20Viterbi math formulation
- 12:40Dynamic programming approach
- 16:37DP benefits visualization
- 18:29Viterbi code walkthrough
- 21:18Speech recognition apps
- 23:04DNA sequencing apps
- 24:51Facial recognition with HMMs
- 27:04Training hidden Markov models introduction
- 28:11Baum-Welch algorithm
- 32:15Q&A
related talks
59mWhen do dependencies in your data help?
23mDynamic inference in probabilistic graphical models
01h 38mIntroduction to Regulatory Genomics and Epigenomics III: Hidden Markov Models in Gene Regulation
57mTutorial on Deep Learning III
01h 07mProbabilistic Programming and Readable Models
42mSome Applications in Human Behavior Modeling
27mTrellis: A Domain-Specific Language for Hidden Markov Models with Sparse Transitions
57mDarwinian Machine Learning: Principles of Machine Learning in Evo-devo, Evo-eco and Evolutionary...