Learning Classifiers for Target Domain with Limited or No Labels

Pengkai Zhu, Hanxiao Wang, Venkatesh Saligrama
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:7643-7653, 2019.

Abstract

In computer vision applications, such as domain adaptation (DA), few shot learning (FSL) and zero-shot learning (ZSL), we encounter new objects and environments, for which insufficient examples exist to allow for training “models from scratch,” and methods that adapt existing models, trained on the presented training environment, to the new scenario are required. We propose a novel visual attribute encoding method that encodes each image as a low-dimensional probability vector composed of prototypical part-type probabilities. The prototypes are learnt to be representative of all training data. At test-time we utilize this encoding as an input to a classifier. At test-time we freeze the encoder and only learn/adapt the classifier component to limited annotated labels in FSL; new semantic attributes in ZSL. We conduct extensive experiments on benchmark datasets. Our method outperforms state-of-art methods trained for the specific contexts (ZSL, FSL, DA).

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-zhu19d, title = {Learning Classifiers for Target Domain with Limited or No Labels}, author = {Zhu, Pengkai and Wang, Hanxiao and Saligrama, Venkatesh}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {7643--7653}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/zhu19d/zhu19d.pdf}, url = {https://proceedings.mlr.press/v97/zhu19d.html}, abstract = {In computer vision applications, such as domain adaptation (DA), few shot learning (FSL) and zero-shot learning (ZSL), we encounter new objects and environments, for which insufficient examples exist to allow for training “models from scratch,” and methods that adapt existing models, trained on the presented training environment, to the new scenario are required. We propose a novel visual attribute encoding method that encodes each image as a low-dimensional probability vector composed of prototypical part-type probabilities. The prototypes are learnt to be representative of all training data. At test-time we utilize this encoding as an input to a classifier. At test-time we freeze the encoder and only learn/adapt the classifier component to limited annotated labels in FSL; new semantic attributes in ZSL. We conduct extensive experiments on benchmark datasets. Our method outperforms state-of-art methods trained for the specific contexts (ZSL, FSL, DA).} }
Endnote
%0 Conference Paper %T Learning Classifiers for Target Domain with Limited or No Labels %A Pengkai Zhu %A Hanxiao Wang %A Venkatesh Saligrama %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-zhu19d %I PMLR %P 7643--7653 %U https://proceedings.mlr.press/v97/zhu19d.html %V 97 %X In computer vision applications, such as domain adaptation (DA), few shot learning (FSL) and zero-shot learning (ZSL), we encounter new objects and environments, for which insufficient examples exist to allow for training “models from scratch,” and methods that adapt existing models, trained on the presented training environment, to the new scenario are required. We propose a novel visual attribute encoding method that encodes each image as a low-dimensional probability vector composed of prototypical part-type probabilities. The prototypes are learnt to be representative of all training data. At test-time we utilize this encoding as an input to a classifier. At test-time we freeze the encoder and only learn/adapt the classifier component to limited annotated labels in FSL; new semantic attributes in ZSL. We conduct extensive experiments on benchmark datasets. Our method outperforms state-of-art methods trained for the specific contexts (ZSL, FSL, DA).
APA
Zhu, P., Wang, H. & Saligrama, V.. (2019). Learning Classifiers for Target Domain with Limited or No Labels. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:7643-7653 Available from https://proceedings.mlr.press/v97/zhu19d.html.

Related Material