TY - GEN
T1 - Scalable training of mixture models via coresets
AU - Feldman, Dan
AU - Faulkner, Matthew
AU - Krause, Andreas
PY - 2011
Y1 - 2011
N2 - How can we train a statistical mixture model on a massive data set? In this paper, we show how to construct coresets for mixtures of Gaussians and natural generalizations. A coreset is a weighted subset of the data, which guarantees that models fitting the coreset will also provide a good fit for the original data set. We show that, perhaps surprisingly, Gaussian mixtures admit coresets of size independent of the size of the data set. More precisely, we prove that a weighted set of O(dκ3/ε2) data points suffices for computing a (1 + ε)-approximation for the optimal model on the original n data points. Moreover, such coresets can be efficiently constructed in a map-reduce style computation, as well as in a streaming setting. Our results rely on a novel reduction of statistical estimation to problems in computational geometry, as well as new complexity results about mixtures of Gaussians. We empirically evaluate our algorithms on several real data sets, including a density estimation problem in the context of earthquake detection using accelerometers in mobile phones.
AB - How can we train a statistical mixture model on a massive data set? In this paper, we show how to construct coresets for mixtures of Gaussians and natural generalizations. A coreset is a weighted subset of the data, which guarantees that models fitting the coreset will also provide a good fit for the original data set. We show that, perhaps surprisingly, Gaussian mixtures admit coresets of size independent of the size of the data set. More precisely, we prove that a weighted set of O(dκ3/ε2) data points suffices for computing a (1 + ε)-approximation for the optimal model on the original n data points. Moreover, such coresets can be efficiently constructed in a map-reduce style computation, as well as in a streaming setting. Our results rely on a novel reduction of statistical estimation to problems in computational geometry, as well as new complexity results about mixtures of Gaussians. We empirically evaluate our algorithms on several real data sets, including a density estimation problem in the context of earthquake detection using accelerometers in mobile phones.
UR - http://www.scopus.com/inward/record.url?scp=85162421236&partnerID=8YFLogxK
M3 - Conference contribution
SN - 9781618395993
T3 - Advances in Neural Information Processing Systems 24: 25th Annual Conference on Neural Information Processing Systems 2011, NIPS 2011
BT - Advances in Neural Information Processing Systems 24
T2 - 25th Annual Conference on Neural Information Processing Systems 2011, NIPS 2011
Y2 - 12 December 2011 through 14 December 2011
ER -