Asymmetric Transfer Learning with Deep Gaussian Processes

Melih Kandemir
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:730-738, 2015.

Abstract

We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by projecting the target data also onto the source domain and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a benchmark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-kandemir15, title = {Asymmetric Transfer Learning with Deep Gaussian Processes}, author = {Kandemir, Melih}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {730--738}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/kandemir15.pdf}, url = {https://proceedings.mlr.press/v37/kandemir15.html}, abstract = {We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by projecting the target data also onto the source domain and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a benchmark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images.} }
Endnote
%0 Conference Paper %T Asymmetric Transfer Learning with Deep Gaussian Processes %A Melih Kandemir %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-kandemir15 %I PMLR %P 730--738 %U https://proceedings.mlr.press/v37/kandemir15.html %V 37 %X We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by projecting the target data also onto the source domain and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a benchmark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images.
RIS
TY - CPAPER TI - Asymmetric Transfer Learning with Deep Gaussian Processes AU - Melih Kandemir BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-kandemir15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 730 EP - 738 L1 - http://proceedings.mlr.press/v37/kandemir15.pdf UR - https://proceedings.mlr.press/v37/kandemir15.html AB - We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by projecting the target data also onto the source domain and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a benchmark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images. ER -
APA
Kandemir, M.. (2015). Asymmetric Transfer Learning with Deep Gaussian Processes. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:730-738 Available from https://proceedings.mlr.press/v37/kandemir15.html.

Related Material