Hidden Markov Model

A Hidden Markov Model (HMM) is a statistical Markov model in which the system being modeled is assumed to be a Markov process with unobserved (hidden) states.

HMM is a generative probabilistic model. HMM is used with data that is represented as a series of states from a series of observations. The transitions between hidden states are assumed to have the form of a (first-order) Markov chain.

HMM is mostly applied in temporal pattern recognition problems.

A real-world example – developing a POS tagger using HMM supervised learning

To demonstrate the HMM algorithm, let's take Part Of Speech (POS) tagging as an example.

POS tagging

POS tagging is important for many Natural Language Processing (NLP) tasks, such as word sense disambiguation, syntax parsing, machine language translation, and Named Entity Recognition (NER). In POS tagging, we identify words as part of speech, such as nouns, verbs, adjectives, and adverbs.

This is a key step in the lowest level of syntactical analysis...