NeurIPS 2016poster14 citations
Active Learning with Oracle Epiphany
Tzu-Kuo Huang, Lihong Li, Ara Vartanian, Saleema Amershi, Xiaojin Zhu
Abstract
We present a theoretical analysis of active learning with more realistic interactions with human oracles. Previous empirical studies have shown oracles abstaining on difficult queries until accumulating enough information to make label decisions. We formalize this phenomenon with an “oracle epiphany model” and analyze active learning query complexity under such oracles for both the realizable and the agnos- tic cases. Our analysis shows that active learning is possible with oracle epiphany, but incurs an additional cost depending on when the epiphany happens. Our results suggest new, principled active learning approaches with realistic oracles.
BibTeX
@inproceedings{NIPS2016_299fb214,
author = {Huang, Tzu-Kuo and Li, Lihong and Vartanian, Ara and Amershi, Saleema and Zhu, Jerry},
booktitle = {Advances in Neural Information Processing Systems},
editor = {D. Lee and M. Sugiyama and U. Luxburg and I. Guyon and R. Garnett},
pages = {},
publisher = {Curran Associates, Inc.},
title = {Active Learning with Oracle Epiphany},
url = {https://proceedings.neurips.cc/paper_files/paper/2016/file/299fb2142d7de959380f91c01c3a293c-Paper.pdf},
volume = {29},
year = {2016}
}