Report on industrial attachment with DSO National laboratories on computational intelligence for knowledge discovery

A Bayesian network is a graph which features conditional probability tables as edges, and variables or events as nodes. This network is a Directed Acyclic Graph the structure reflects the dependencies of the nodes. There are several algorithms available to learn a Bayesian network, and the...

Full description

Saved in:
Bibliographic Details
Main Author: Kok, Hong Jie
Other Authors: Sui Qing
Format: Industrial Attachment (IA)
Language:English
Published: 2015
Subjects:
Online Access:http://hdl.handle.net/10356/65252
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:A Bayesian network is a graph which features conditional probability tables as edges, and variables or events as nodes. This network is a Directed Acyclic Graph the structure reflects the dependencies of the nodes. There are several algorithms available to learn a Bayesian network, and the focus here is on latent tree learning algorithms which can discover structures with hidden nodes, which may reflect simpler relationships and better categorization of data. By integrating these algorithms into an existing learning knowledge system, the evaluation of performance in terms of structure scoring metrics and classification accuracy can be carried out to compare the effectiveness of these algorithms to those traditional learning methods.