Task Understanding from Confusing Multi-task Data

Xin Su, Yizhou Jiang, Shangqi Guo, Feng Chen
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:9177-9186, 2020.

Abstract

Beyond machine learning’s success in the specific tasks, research for learning multiple tasks simultaneously is referred to as multi-task learning. However, existing multi-task learning needs manual definition of tasks and manual task annotation. A crucial problem for advanced intelligence is how to understand the human task concept using basic input-output pairs. Without task definition, samples from multiple tasks are mixed together and result in a confusing mapping challenge. We propose Confusing Supervised Learning (CSL) that takes these confusing samples and extracts task concepts by differentiating between these samples. We theoretically proved the feasibility of the CSL framework and designed an iterative algorithm to distinguish between tasks. The experiments demonstrate that our CSL methods could achieve a human-like task understanding without task labeling in multi-function regression problems and multi-task recognition problems.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-su20b, title = {Task Understanding from Confusing Multi-task Data}, author = {Su, Xin and Jiang, Yizhou and Guo, Shangqi and Chen, Feng}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {9177--9186}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/su20b/su20b.pdf}, url = {https://proceedings.mlr.press/v119/su20b.html}, abstract = {Beyond machine learning’s success in the specific tasks, research for learning multiple tasks simultaneously is referred to as multi-task learning. However, existing multi-task learning needs manual definition of tasks and manual task annotation. A crucial problem for advanced intelligence is how to understand the human task concept using basic input-output pairs. Without task definition, samples from multiple tasks are mixed together and result in a confusing mapping challenge. We propose Confusing Supervised Learning (CSL) that takes these confusing samples and extracts task concepts by differentiating between these samples. We theoretically proved the feasibility of the CSL framework and designed an iterative algorithm to distinguish between tasks. The experiments demonstrate that our CSL methods could achieve a human-like task understanding without task labeling in multi-function regression problems and multi-task recognition problems.} }
Endnote
%0 Conference Paper %T Task Understanding from Confusing Multi-task Data %A Xin Su %A Yizhou Jiang %A Shangqi Guo %A Feng Chen %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-su20b %I PMLR %P 9177--9186 %U https://proceedings.mlr.press/v119/su20b.html %V 119 %X Beyond machine learning’s success in the specific tasks, research for learning multiple tasks simultaneously is referred to as multi-task learning. However, existing multi-task learning needs manual definition of tasks and manual task annotation. A crucial problem for advanced intelligence is how to understand the human task concept using basic input-output pairs. Without task definition, samples from multiple tasks are mixed together and result in a confusing mapping challenge. We propose Confusing Supervised Learning (CSL) that takes these confusing samples and extracts task concepts by differentiating between these samples. We theoretically proved the feasibility of the CSL framework and designed an iterative algorithm to distinguish between tasks. The experiments demonstrate that our CSL methods could achieve a human-like task understanding without task labeling in multi-function regression problems and multi-task recognition problems.
APA
Su, X., Jiang, Y., Guo, S. & Chen, F.. (2020). Task Understanding from Confusing Multi-task Data. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:9177-9186 Available from https://proceedings.mlr.press/v119/su20b.html.

Related Material