arXiv subjects

Yuanyuan Lin

Publications and source records attributed to Yuanyuan Lin.

22 records · Page 2Linked to original sources

arXiv2021

Non-asymptotic Excess Risk Bounds for Classification with Deep Convolutional Neural Networks

In this paper, we consider the problem of binary classification with a class of general deep convolutional neural networks, which includes fully-connected neural networks and fully convolutional neural networks as special cases. We establish non-asymptotic excess risk bounds for a class of convex surrogate losses and target functions with different modulus of continuity. An important feature of our results is that we clearly define the prefactors of the risk bounds in terms of the input data dimension and other model parameters and show that they depend polynomially on the dimensionality in some important models. We also show that the classification methods with CNNs can circumvent the curse of dimensionality if the input data is supported on an approximate low-dimensional manifold. To establish these results, we derive an upper bound for the covering number for the class of general convolutional neural networks with a bias term in each convolutional layer, and derive new results on the approximation power of CNNs for any uniformly-continuous target functions. These results provide further insights into the complexity and the approximation power of general convolutional neural networks, which are of independent interest and may have other applications. Finally, we apply our general results to analyze the non-asymptotic excess risk bounds for four widely used methods with different loss functions using CNNs, including the least squares, the logistic, the exponential and the SVM hinge losses.

cs.LG↗

arXiv2017

Nearly Semiparametric Efficient Estimation of Quantile Regression

As a competitive alternative to least squares regression, quantile regression is popular in analyzing heterogenous data. For quantile regression model specified for one single quantile level $τ$, major difficulties of semiparametric efficient estimation are the unavailability of a parametric efficient score and the conditional density estimation. In this paper, with the help of the least favorable submodel technique, we first derive the semiparametric efficient scores for linear quantile regression models that are assumed for a single quantile level, multiple quantile levels and all the quantile levels in $(0,1)$ respectively. Our main discovery is a one-step (nearly) semiparametric efficient estimation for the regression coefficients of the quantile regression models assumed for multiple quantile levels, which has several advantages: it could be regarded as an optimal way to pool information across multiple/other quantiles for efficiency gain; it is computationally feasible and easy to implement, as the initial estimator is easily available; due to the nature of quantile regression models under investigation, the conditional density estimation is straightforward by plugging in an initial estimator. The resulting estimator is proved to achieve the corresponding semiparametric efficiency lower bound under regularity conditions. Numerical studies including simulations and an example of birth weight of children confirms that the proposed estimator leads to higher efficiency compared with the Koenker-Bassett quantile regression estimator for all quantiles of interest.

stat.ME↗

arXiv2014

Regression Analysis with Response-biased Sampling

Response-biased sampling, in which samples are drawn from a popula- tion according to the values of the response variable, is common in biomedical, epidemiological, economic and social studies. In particular, the complete obser- vations in data with censoring, truncation or missing covariates can be regarded as response-biased sampling under certain conditions. This paper proposes to use transformation models, known as the generalized accelerated failure time model in econometrics, for regression analysis with response-biased sampling. With unknown error distribution, the transformation models are broad enough to cover linear re- gression models, the Cox's model and the proportional odds model as special cases. To the best of our knowledge, except for the case-control logistic regression, there is no report in the literature that a prospective estimation approach can work for biased sampling without any modification. We prove that the maximum rank corre- lation estimation is valid for response-biased sampling and establish its consistency and asymptotic normality. Unlike the inverse probability methods, the proposed method of estimation does not involve the sampling probabilities, which are often difficult to obtain in practice. Without the need of estimating the unknown trans- formation function or the error distribution, the proposed method is numerically easy to implement with the Nelder-Mead simplex algorithm, which does not require convexity or continuity. We propose an inference procedure using random weight- ing to avoid the complication of density estimation when using the plug-in rule for variance estimation. Numerical studies with supportive evidence are presented. Applications are illustrated with the Forbes Global 2000 data and the Stanford heart transplant data.

stat.ME↗

arXiv2013

Least Product Relative Error Estimation

A least product relative error criterion is proposed for multiplicative regression models. It is invariant under scale transformation of the outcome and covariates. In addition, the objective function is smooth and convex, resulting in a simple and uniquely defined estimator of the regression parameter. It is shown that the estimator is asymptotically normal and that the simple plugging-in variance estimation is valid. Simulation results confirm that the proposed method performs well. An application to body fat calculation is presented to illustrate the new method.

stat.ME↗