Fetching the paper…

Reading the bibliography…

2008

Importance Weighted Active Learning

Beygelzimer, Alina, Dasgupta, Sanjoy, Langford, John

Original
PDF
Post on X

Understand

We present a practical and statistically consistent scheme for actively learning binary classifiers under general loss functions.

Our algorithm uses importance weighting to correct sampling bias, and by controlling the variance, we are able to give rigorous label complexity bounds for the learning process.
Experiments on passively labeled data show that this approach reduces the label complexity required to achieve good predictive performance on many learning problems.

Importance Weighted Active Learning · Around

Built on

Weighted sums of certain dependent random variables
K. Azuma · 1967
Earlier work this paper cites.
Distribution inequalities for the binomial law
E. Slud · 1977
Earlier work this paper cites.
Improving generalization with active learning
D. Cohn, L. Atlas, and R. Ladner · 1994
Earlier work this paper cites.
A Probabilistic Theory of Pattern Recognition
L. Devroye, L. Gyorfi, and G. Lugosi · 1996
Earlier work this paper cites.
Query learning strategies using boosting and bagging
N. Abe and H. Mamitsuka · 1998
Earlier work this paper cites.
Cost-sensitive learning by cost-proportionate example weighting
B. Zadrozny, J. Langford, and N. Abe · 2003
Earlier work this paper cites.

Similar

Convex Optimization
S. Boyd and L Vandenberghe · 2004
Cited alongside, same era.
Analysis of perceptron-based active learning
S. Dasgupta, A. Tauman Kalai, and C. Monteleoni · 2005
Cited alongside, same era.
Practical prediction theory for classification
J. Langford · 2005
Cited alongside, same era.
Agnostic active learning
M.-F. Balcan, A. Beygelzimer, and J. Langford · 2006
Cited alongside, same era.
Active learning in the non-realizable case
M. Kääriäinen · 2006
Cited alongside, same era.
Optimal Design of Experiments , volume 50 of Classics in Applied Mathematics
F. Pukelsheim · 2006
Cited alongside, same era.

Then

Active learning for misspecified models
M. Sugiyama · 2006
Later among the works it cites.
Active learning for misspecified generalized linear models
F. Bach · 2007
Later among the works it cites.
A bound on the label complexity of agnostic active learning
S. Hanneke · 2007
Later among the works it cites.
Hierarchical sampling for active learning
S. Dasgupta and D. Hsu · 2008
Closest in time.
A general agnostic active learning algorithm
S. Dasgupta, D. Hsu, and C. Monteleoni · 2008
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…