# HMM & decoding

Subject: Natural Language Processing

Exam: TEE

Reader: https://deep-learning-study-room.vercel.app/nlp/#tee/formulas/f-hmm

Canonical: https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas-chapter/f-hmm/

Formulas explained through full worked examples

1. [Markov chains: read a transition table and multiply a path](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--markov-chain/)
2. [Estimate transition and emission probabilities from tagged data](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--transition-emission-estimation/)
3. [HMM path probabilities: three complete numerical examples](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--hot-cold-paths/)
4. [POS HMM worked corpus: Mary, Jane, can, will and spot](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--tagging-corpus-case/)
5. [Why exhaustive search grows exponentially](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--sequence-count/)
6. [Viterbi decoding: compare incoming paths, keep the best, trace back](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--viterbi-explained/)
7. [A path score is not the probability of the observed sentence](https://deep-learning-study-room.vercel.app/read/nlp/tee/formulas/f-hmm--joint-vs-posterior/)