A New Modeling for Knowledge Transfer in Machine Learning: Minimum Enclosing Ball-based Learner Independent Knowledge Transfer for Correlated Multi-task Learning - Couverture souple

Liu, Fan

 
9783844397321: A New Modeling for Knowledge Transfer in Machine Learning: Minimum Enclosing Ball-based Learner Independent Knowledge Transfer for Correlated Multi-task Learning
Couverture souple
ISBN 10 : 3844397329 ISBN 13 : 9783844397321
Editeur : LAP LAMBERT Academic Publishing, 2011

Synopsis

Multi-Task Learning (MTL), as opposed to Single Task Learning (STL), has become a hot topic in machine learning research. MTL has shown significant advantage to STL because of its ability to facilitate knowledge sharing between tasks. This thesis presents my recent studies on Knowledge Transfer (KT) – the process of transferring knowledge from one task to another, which is at the core of MTL. The novelly proposed KT algorithm for correlated MTL adapts learner independence, thus empowering any ordinary classifier for MTL. The proposed MEB-based KT is on the basis that in the feature space, the two correlated tasks share some common input data that lie on the overlapping regions of the feature spaces in-between the two correlated tasks. The main idea is to find the correlating knowledge – overlapping regions of the two tasks – and transfer the related data regardless of the learner employed. KT is done by building a correlation space via MEBs and transferring the enclosed instances from the primary task to the secondary task. The extent of KT depends on the amount of overlapping instances between two tasks. This book is required reading for post-graduates and researchers in MTL.

Les informations fournies dans la section « Synopsis » peuvent faire référence à une autre édition de ce titre.

Présentation de l'éditeur

Multi-Task Learning (MTL), as opposed to Single Task Learning (STL), has become a hot topic in machine learning research. MTL has shown significant advantage to STL because of its ability to facilitate knowledge sharing between tasks. This thesis presents my recent studies on Knowledge Transfer (KT) – the process of transferring knowledge from one task to another, which is at the core of MTL. The novelly proposed KT algorithm for correlated MTL adapts learner independence, thus empowering any ordinary classifier for MTL. The proposed MEB-based KT is on the basis that in the feature space, the two correlated tasks share some common input data that lie on the overlapping regions of the feature spaces in-between the two correlated tasks. The main idea is to find the correlating knowledge – overlapping regions of the two tasks – and transfer the related data regardless of the learner employed. KT is done by building a correlation space via MEBs and transferring the enclosed instances from the primary task to the secondary task. The extent of KT depends on the amount of overlapping instances between two tasks. This book is required reading for post-graduates and researchers in MTL.

Biographie de l'auteur

Fan Liu has received a MCIS at AUT, New Zealand, in 2011. The MCIS research focuses on Co-Learning Multi-Task Pattern Recognition using Minimum Enclosing Balls. Fan has been awarded a scholarship in connection with the NICT Project that KEDRI is involved with. Currently, Fan is a Doctoral Candidate and interested in Swarm Intelligence for Robotics.

Les informations fournies dans la section « A propos du livre » peuvent faire référence à une autre édition de ce titre.

Éditeur
LAP LAMBERT Academic Publishing
Date d'édition
2011
Langue
anglais
ISBN 10 
3844397329
ISBN 13 
9783844397321
Reliure
Broché
Nombre de pages
88
Coordonnées du fabricant
non disponible
Personne responsable
KreGa Onlinehandel GmbH
4950000000000
info@krega-onlinehandel.de

Birkenweg 31
Brakel
33034
Allemagne