A Precise High-Dimensional Asymptotic Theory for Boosting and Minimum-L1-Norm Interpolated Classifiers
A Precise High-Dimensional Asymptotic Theory for Boosting and Minimum-L1-Norm Interpolated Classifiers
复制标题
Boosting 和最小 L1 范数插值分类器的精确高维渐近理论
DOI:
--
复制
发表时间:
2020
期刊:
影响因子:
--
通讯作者:
P. Sur
中科院分区:
文献类型:
--
作者:
Tengyuan Liang;P. Sur
This paper establishes a precise high-dimensional asymptotic theory for boosting on separable data, taking statistical and computational perspectives. We consider a high-dimensional setting where the number of features (weak learners) $p$ scales with the sample size $n$, in an overparametrized regime. Under a class of statistical models, we provide an exact analysis of the generalization error of boosting when the algorithm interpolates the training data and maximizes the empirical $\ell_1$-margin. Further, we explicitly pin down the relation between the boosting test error and the optimal Bayes error, as well as the proportion of active features at interpolation (with zero initialization). In turn, these precise characterizations answer certain questions raised in \cite{breiman1999prediction, schapire1998boosting} surrounding boosting, under assumed data generating processes. At the heart of our theory lies an in-depth study of the maximum-$\ell_1$-margin, which can be accurately described by a new system of non-linear equations; to analyze this margin, we rely on Gaussian comparison techniques and develop a novel uniform deviation argument. Our statistical and computational arguments can handle (1) any finite-rank spiked covariance model for the feature distribution and (2) variants of boosting corresponding to general $\ell_q$-geometry, $q \in [1, 2]$. As a final component, via the Lindeberg principle, we establish a universality result showcasing that the scaled $\ell_1$-margin (asymptotically) remains the same, whether the covariates used for boosting arise from a non-linear random feature model or an appropriately linearized model with matching moments.
登录
查看更多内容
DOI:
10.1080/01621459.2020.1745812
发表时间:
2019-01
影响因子:
3.7
作者:
Xialiang Dou;Tengyuan Liang
通讯作者:
Xialiang Dou;Tengyuan Liang
DOI:
--
发表时间:
2019
期刊:
IEEE International Symposium on Information Theory
影响因子:
--
作者:
Hu, Hong;Lu, Yue M.
通讯作者:
Lu, Yue M.
DOI:
10.1080/01621459.2016.1273116
发表时间:
2015-10
影响因子:
3.7
作者:
Alexander Hanbo Li;Jelena Bradic
通讯作者:
Alexander Hanbo Li;Jelena Bradic
影响因子:
2
作者:
P. Sur;Yuxin Chen;E. Candès
通讯作者:
P. Sur;Yuxin Chen;E. Candès
DOI:
--
发表时间:
2018-06
期刊:
ArXiv
影响因子:
--
作者:
M. Belkin;A. Rakhlin;A. Tsybakov
通讯作者:
M. Belkin;A. Rakhlin;A. Tsybakov