# Learning to Learn with the Informative Vector Machine

Neil D. Lawrence, University of Sheffield
John C. Platt, Google

in Proceedings of the International Conference in Machine Learning 21, pp 512-519

#### Abstract

This paper describes an efficient method for learning the parameters of a Gaussian process (GP). The parameters are learned from multiple tasks which are assumed to have been drawn independently from the same GP prior. An efficient algorithm is obtained by extending the informative vector machine (IVM) algorithm to handle the multi-task learning case. The multi-task IVM (MT-IVM) saves computation by greedily selecting the most informative examples from the separate tasks. The MT-IVM is also shown to be more efficient than sub-sampling on an artificial data-set and more effective than the traditional IVM in a speaker dependent phoneme recognition task.

  @InProceedings{lawrence-learning04, title = {Learning to Learn with the Informative Vector Machine}, author = {Neil D. Lawrence and John C. Platt}, booktitle = {Proceedings of the International Conference in Machine Learning}, pages = {512}, year = {2004}, editor = {Russell Greiner and Dale Schuurmans}, volume = {21}, month = {00}, publisher = {Omnipress}, edit = {https://github.com/lawrennd//publications/edit/gh-pages/_posts/2004-01-01-lawrence-learning04.md}, url = {http://inverseprobability.com/publications/lawrence-learning04.html}, abstract = {This paper describes an efficient method for learning the parameters of a Gaussian process (GP). The parameters are learned from multiple tasks which are assumed to have been drawn independently from the same GP prior. An efficient algorithm is obtained by extending the informative vector machine (IVM) algorithm to handle the multi-task learning case. The multi-task IVM (MT-IVM) saves computation by greedily selecting the most informative examples from the separate tasks. The MT-IVM is also shown to be more efficient than sub-sampling on an artificial data-set and more effective than the traditional IVM in a speaker dependent phoneme recognition task.}, crossref = {Greiner:icml04}, key = {Lawrence:learning04}, doi = {10.1145/1015330.1015382}, linkpdf = {ftp://ftp.dcs.shef.ac.uk/home/neil/mtivm.pdf}, linkpsgz = {ftp://ftp.dcs.shef.ac.uk/home/neil/mtivm.ps.gz}, linksoftware = {http://inverseprobability.com/mtivm/}, group = {shefml,gp,spgp} }
 %T Learning to Learn with the Informative Vector Machine %A Neil D. Lawrence and John C. Platt %B %C Proceedings of the International Conference in Machine Learning %D %E Russell Greiner and Dale Schuurmans %F lawrence-learning04 %I Omnipress %P 512--519 %R 10.1145/1015330.1015382 %U http://inverseprobability.com/publications/lawrence-learning04.html %V 21 %X This paper describes an efficient method for learning the parameters of a Gaussian process (GP). The parameters are learned from multiple tasks which are assumed to have been drawn independently from the same GP prior. An efficient algorithm is obtained by extending the informative vector machine (IVM) algorithm to handle the multi-task learning case. The multi-task IVM (MT-IVM) saves computation by greedily selecting the most informative examples from the separate tasks. The MT-IVM is also shown to be more efficient than sub-sampling on an artificial data-set and more effective than the traditional IVM in a speaker dependent phoneme recognition task. 
 TY - CPAPER TI - Learning to Learn with the Informative Vector Machine AU - Neil D. Lawrence AU - John C. Platt BT - Proceedings of the International Conference in Machine Learning PY - 2004/01/01 DA - 2004/01/01 ED - Russell Greiner ED - Dale Schuurmans ID - lawrence-learning04 PB - Omnipress SP - 512 EP - 519 DO - 10.1145/1015330.1015382 L1 - ftp://ftp.dcs.shef.ac.uk/home/neil/mtivm.pdf UR - http://inverseprobability.com/publications/lawrence-learning04.html AB - This paper describes an efficient method for learning the parameters of a Gaussian process (GP). The parameters are learned from multiple tasks which are assumed to have been drawn independently from the same GP prior. An efficient algorithm is obtained by extending the informative vector machine (IVM) algorithm to handle the multi-task learning case. The multi-task IVM (MT-IVM) saves computation by greedily selecting the most informative examples from the separate tasks. The MT-IVM is also shown to be more efficient than sub-sampling on an artificial data-set and more effective than the traditional IVM in a speaker dependent phoneme recognition task. ER - 
 Lawrence, N.D. & Platt, J.C.. (2004). Learning to Learn with the Informative Vector Machine. Proceedings of the International Conference in Machine Learning 21:512-519