000234521 001__ 234521
000234521 005__ 20190317000916.0
000234521 037__ $$aCONF
000234521 245__ $$aLearning Active Learning from Data
000234521 269__ $$a2017
000234521 260__ $$c2017
000234521 336__ $$aConference Papers
000234521 520__ $$aIn this paper, we suggest a novel data-driven approach to active learning (AL). The key idea is to train a regressor that predicts the expected error reduction for a candidate sample in a particular learning state. By formulating the query selection procedure as a regression problem we are not restricted to working with existing AL heuristics; instead, we learn strategies based on experience from previous AL outcomes. We show that a strategy can be learnt either from simple synthetic 2D datasets or from a subset of domain-specific data. Our method yields strategies that work well on real data from a wide range of domains.
000234521 6531_ $$aactive learning
000234521 6531_ $$ameta-learning
000234521 700__ $$g237229$$aKonyushkova, Ksenia$$0247782
000234521 700__ $$aSznitman, Raphael
000234521 700__ $$aFua, Pascal
000234521 7112_ $$aAdvances in Neural Information Processing Systems
000234521 7112_ $$aConference on Neural Information Processing Systems (NIPS)
000234521 8560_ $$fpascal.fua@epfl.ch
000234521 8564_ $$uhttp://papers.nips.cc/paper/7010-learning-active-learning-from-data.pdf$$zURL
000234521 8564_ $$uhttps://infoscience.epfl.ch/record/234521/files/LAL.pdf$$zPublisher's version$$s815460$$yPublisher's version
000234521 909C0 $$xU10659$$0252087$$pCVLAB
000234521 909CO $$ooai:infoscience.tind.io:234521$$qGLOBAL_SET$$pconf$$pIC
000234521 917Z8 $$x237229
000234521 937__ $$aEPFL-CONF-234521
000234521 973__ $$rREVIEWED$$aEPFL
000234521 980__ $$aCONF