Continuous-time baum-welch training

Patent metadata

Field	Value
Publication number	US-9508045-B2
Application number	US-201213588912-A
Country	US
Kind code	B2
Filing date	Aug 17, 2012
Priority date	Aug 17, 2012
Publication date	Nov 29, 2016
Grant date	Nov 29, 2016

Abstract

The apparatus, systems, and methods described herein may operate to receive information identifying and describing at least one of a set of events, an initial distribution of a plurality of states, an initial transition matrix, or an initial event matrix; generate, based at least in part on the information, at least one intermediate transition matrix and at least one intermediate event matrix describing a sparse Baum-Welch training that allows no event to occur at one or more time steps; and transform the at least one intermediate transition matrix and the at least one intermediate event matrix into a transition matrix and an event matrix describing a continuous-time Baum-Welch training, the continuous-time Baum-Welch training allowing events to occur simultaneously or at sporadic time intervals in a Markov model including a hidden Markov Model (HMM) having more than two hidden states.

First claim

What is claimed is:

1. A method comprising:
receiving, at a data training module executable by one or more hardware processors of a training node and from a source node, information identifying and describing at least one of a set of events, an initial distribution of a plurality of states, an initial transition matrix, or an initial event matrix;
generating, at the data training module and based at least in part on the information, at least one intermediate transition matrix an…

Assignees

• Fisher David Charles
• Raytheon Co

Inventors

• Fisher David Charles

Classifications

• G06N7/01

  Physics · mapped topic

• G06N99/005 Primary

  Physics · mapped topic

• G06N7/005

  Physics · mapped topic

• G06N20/00 Primary

  Physics · mapped topic