https://github.com/adrianulbona/hmm

Hidden Markov Models Java Library
https://github.com/adrianulbona/hmm

Last synced: 10 days ago
JSON representation

Hidden Markov Models Java Library

• Host: GitHub
• URL: https://github.com/adrianulbona/hmm
• Owner: adrianulbona
• License: apache-2.0
• Created: 2016-01-13T03:48:08.000Z (almost 9 years ago)
• Default Branch: master
• Last Pushed: 2016-12-20T07:57:08.000Z (almost 8 years ago)
• Last Synced: 2024-10-13T00:10:55.441Z (25 days ago)
• Language: Java
• Homepage: http://adrianulbona.github.io/hmm/
• Size: 54.7 KB
• Stars: 40
• Watchers: 8
• Forks: 19
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## HMM abstractions in Java 8 

[![Build Status](https://travis-ci.org/adrianulbona/hmm.svg)](https://travis-ci.org/adrianulbona/hmm)

[![Maven Central](https://maven-badges.herokuapp.com/maven-central/io.github.adrianulbona/hmm/badge.svg)](https://maven-badges.herokuapp.com/maven-central/io.github.adrianulbona/hmm)

Besides the basic abstractions, a most probable state sequence solution is implemented based on the Viterbi algorithm.

### The library is hosted on Maven Central:

#### Maven

```xml

    io.github.adrianulbona

    hmm

    0.1.0

```

#### Gradle

```groovy

compile 'io.github.adrianulbona:hmm:0.1.0'

```

### How to use it:

#### Getting the most probable sequence of states based on a sequence of observations:

```java

Model model = WikipediaViterbi.INSTANCE.model;

List symptoms = asList(NORMAL, COLD, DIZZY);

List evolution = new MostProbableStateSequenceFinder<>(model).basedOn(symptoms);

```

#### How to define a model: 

```java

public enum WikipediaViterbi {

	INSTANCE;

	public final Model model;

	WikipediaViterbi() {

		model = new Model<>(probabilityCalculator(), reachableStatesFinder());

	}

	public enum MedicalState implements State {

		HEALTHY,

		FEVER;

	}

	

	public enum Symptom implements Observation {

		NORMAL,

		COLD,

		DIZZY;

	}

	

	private ProbabilityCalculator probabilityCalculator() {

		return new ProbabilityCalculator<>(StartProbabilities.INSTANCE.data::get,

				EmissionProbabilities.INSTANCE.data::get, 

				TransitionProbabilities.INSTANCE.data::get);

	}

	private ReachableStateFinder reachableStatesFinder() {

		return observation -> asList(MedicalState.values());

	}

	private enum StartProbabilities {

		INSTANCE;

		public final Map data;

		StartProbabilities() {

			data = new HashMap<>();

			data.put(MedicalState.HEALTHY, 0.6);

			data.put(MedicalState.FEVER, 0.4);

		}

	}

	private enum TransitionProbabilities {

		INSTANCE;

		public final Map, Double> data;

		TransitionProbabilities() {

			data = new HashMap<>();

			data.put(new Transition<>(MedicalState.HEALTHY, MedicalState.HEALTHY), 0.7);

			data.put(new Transition<>(MedicalState.HEALTHY, MedicalState.FEVER), 0.3);

			data.put(new Transition<>(MedicalState.FEVER, MedicalState.HEALTHY), 0.4);

			data.put(new Transition<>(MedicalState.FEVER, MedicalState.FEVER), 0.6);

		}

	}

	private enum EmissionProbabilities {

		INSTANCE;

		public final Map, Double> data;

		EmissionProbabilities() {

			data = new HashMap<>();

			data.put(new Emission<>(MedicalState.HEALTHY, Symptom.NORMAL), 0.5);

			data.put(new Emission<>(MedicalState.HEALTHY, Symptom.COLD), 0.4);

			data.put(new Emission<>(MedicalState.HEALTHY, Symptom.DIZZY), 0.1);

			data.put(new Emission<>(MedicalState.FEVER, Symptom.NORMAL), 0.1);

			data.put(new Emission<>(MedicalState.FEVER, Symptom.COLD), 0.3);

			data.put(new Emission<>(MedicalState.FEVER, Symptom.DIZZY), 0.6);

		}

	}

}

```