Online Estimation of the Average Treatment Effect 437
setting, cross-validation can be used to select between many estimators, or a model stacking
approach such as the super learner algorithm [17] can be used to choose a combination of
estimators. A similar approach can be taken in an online setting by using each new minibatch
as an independent validation set to estimate out-of-sample performance before updating an
estimator with that minibatch. In this way, many estimators with different choices of tuning
parameters can be fit concurrently, and the best or a combination can be selected based on
the estimated out-of-sample performance.
References
1. Hanady Abdulsalam, David B. Skillicorn, and Patrick Martin. Streaming random
forests. In Database Engineering and Applications Symposium, IDEAS 2007. 11th
International, pp. 225–232. IEEE, Alberta, Canada, 2007.
2. Peter J. Bickel, Chris A.J. Klaassen, Ya’acov Ritov, and Jon A. Wellner. Efficient
and Adaptive Estimation for Semiparametric Models. Johns Hopkins University Press,
Baltimore, MD, 1993.
3. Antoine Bordes, L´eon Bottou, and Patrick Gallinari. SGD-QN: Careful quasi-Newton
stochastic gradient descent. The Journal of Machine Learning Research, 10:1737–1754,
2009.
4. L´eon Bottou. Large-scale machine learning with stochastic gradient descent. In
Proceedings of COMPSTAT, pp. 177–186. Springer, Berlin, Germany, 2010.
5. L´eon Bottou. Stochastic gradient descent tricks. In Gr´egoire Montavon; Genevi`eve B.
Orr; Klaus-Robert M¨uller, editors, Neural Networks: Tricks of the Trade, pp. 421–436.
Springer, Berlin, Germany, 2012.
6. L´eon Bottou and Yann L. Cun. On-line learning for very large data sets. Applied
Stochastic Models in Business and Industry, 21(2):137–151, 2005.
7. John Duchi, Elad Hazan, and Yoram Singer. Adaptive subgradient methods for online
learning and stochastic optimization. The Journal of Machine Learning Research,
12:2121–2159, 2011.
8. Noboru Murata. A Statistical Study of On-Line Learning: Online Learning and Neural
Networks. Cambridge University Press, Cambridge, 1998.
9. Nikunj C. Oza. Online bagging and boosting. In IEEE International Conference on
Systems, Man and Cybernetics, volume 3, pp. 2340–2345. IEEE, 2005.
10. Boris T. Polyak and Anatoli B. Juditsky. Acceleration of stochastic approximation by
averaging. SIAM Journal on Control and Optimization, 30(4):838–855, 1992.
11. James M. Robins, Andrea Rotnitzky, and Lue P. Zhao. Estimation of regression
coefficients when some regressors are not always observed. Journal of the American
Statistical Association, 89(427):846–866, 1994.
12. Paul R. Rosenbaum and Donald B. Rubin. The central role of the propensity score in
observational studies for causal effects. Biometrika, 70(1):41–55, 1983.