Part of Advances in Neural Information Processing Systems 29 (NIPS 2016)
Tzu-Kuo Huang, Lihong Li, Ara Vartanian, Saleema Amershi, Jerry Zhu
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.