On Discarding, Caching, and Recalling Samples in Active Learning
Ashish Kapoor, Eric Horvitz
We address challenges of active learning un- der scarce informational resources in non- stationary environments. In real-world set- tings, data labeled and integrated into a predictive model may become invalid over time. However, the data can become infor- mative again with switches in context and such changes may indicate unmodeled cyclic or other temporal dynamics. We explore principles for discarding, caching, and re- calling labeled data points in active learn- ing based on computations of value of infor- mation. We review key concepts and study the value of the methods via investigations of predictive performance and costs of acquiring data for simulated and real-world data sets.
PDF Link: /papers/07/p209-kapoor.pdf
AUTHOR = "Ashish Kapoor
and Eric Horvitz",
TITLE = "On Discarding, Caching, and Recalling Samples in Active Learning",
BOOKTITLE = "Proceedings of the Twenty-Third Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-07)",
PUBLISHER = "AUAI Press",
ADDRESS = "Corvallis, Oregon",
YEAR = "2007",
PAGES = "209--216"