北航数学论坛学术报告

Leveraging unlabeled data and model averaging to improve prediction

张新雨（中科院数学与系统科学研究院）

时间：2024年11月27日（周三）上午10:00-11:00

地点：沙河国实E404

摘要: The available data in semi-supervised learning usually consists of relatively small sized labeled data and much larger sized unlabeled data. How to effectively exploit unlabeled data is the key issue. In this paper, we consider the prediction of the regression function. Write regression function in the form of a copula and marginal distributions, and the unlabeled data can be exploited to improve the estimation of the marginal distributions. The predictions of the different copulas are weighted, where the weights are obtained by minimizing an estimate of the risk. Error-ambiguity decomposition of the risk is performed such that unlabeled data can be exploited to improve the estimation of the risk. We demonstrate the asymptotic normality of copula parameters and regression function estimates of the candidate model under the semi-supervised framework, as well as the asymptotic optimality and weight consistency of model averaging estimates. Our model averaging estimate achieves faster convergence rates of asymptotic optimality and weight consistency than the supervised counterpart. Extensive simulation experiments demonstrate the effectiveness of the proposed method.

报告人简介:

张新雨，中科院数学与系统科学研究院研究员。主要从事统计学的理论和应用研究工作，具体研究方向包括模型平均、机器学习、经济预测、医学统计等，担任国内SCI期刊《Journal of Systems Science & Complexity (JSSC)》领域主编和其他5个国内外重要期刊的编委。先后主持国家级青年人才项目和国家级人才项目。

邀请人：夏勇

欢迎大家参加！