Flexible latent variable models for multi-task learning

Authors:
Jian Zhang;Zoubin Ghahramani;Yiming Yang
Affiliations:
Department of Statistics, Purdue University, West Lafayette, USA 47907;Department of Engineering, University of Cambridge, Cambridge, UK CB2 1PZ and School of Computer Science, Carnegie Mellon University, Pittsburgh, USA 15213;School of Computer Science, Carnegie Mellon University, Pittsburgh, USA 15213
Venue:
Machine Learning
Year:
2008

Citing 7
Cited 8

Multitask Learning

Machine Learning - Special issue on inductive transfer
Learning to learn

Learning to learn
Hierarchically Classifying Documents Using Very Few Words

ICML '97 Proceedings of the Fourteenth International Conference on Machine Learning
Empirical Bayes for Learning to Learn

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Learning Multiple Tasks with Kernel Methods

The Journal of Machine Learning Research
Learning Gaussian processes from multiple tasks

ICML '05 Proceedings of the 22nd international conference on Machine learning
A model of inductive bias learning

Journal of Artificial Intelligence Research

Guest editor's introduction: special issue on inductive transfer learning

Machine Learning
Automatic Choice of Control Measurements

ACML '09 Proceedings of the 1st Asian Conference on Machine Learning: Advances in Machine Learning
Multi-task feature learning via efficient l2, 1-norm minimization

UAI '09 Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence
Discriminative factored prior models for personalized content-based recommendation

CIKM '10 Proceedings of the 19th ACM international conference on Information and knowledge management
Relevant subtask learning by constrained mixture models

Intelligent Data Analysis
Drosophila Gene Expression Pattern Annotation through Multi-Instance Multi-Label Learning

IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)
Learning output kernels for multi-task problems

Neurocomputing
Geometry preserving multi-task metric learning

Machine Learning

Quantified Score

Hi-index	0.00

Visualization

Abstract

Given multiple prediction problems such as regression or classification, we are interested in a joint inference framework that can effectively share information between tasks to improve the prediction accuracy, especially when the number of training examples per problem is small. In this paper we propose a probabilistic framework which can support a set of latent variable models for different multi-task learning scenarios. We show that the framework is a generalization of standard learning methods for single prediction problems and it can effectively model the shared structure among different prediction tasks. Furthermore, we present efficient algorithms for the empirical Bayes method as well as point estimation. Our experiments on both simulated datasets and real world classification datasets show the effectiveness of the proposed models in two evaluation settings: a standard multi-task learning setting and a transfer learning setting.