报告题目：Kernel Variable Importance Measure and Its Application

报告时间：2023-09-21   14:00-16:00

报告人：彭柳华  副教授  墨尔本大学

腾讯会议ID：139-475-473

会议链接：https://meeting.tencent.com/dm/7JXqW072sPd8

Abstract: This project proposes kernel variable importance measure (KIM) based on the well-known maximum mean discrepancy (MMD). The KIM can effectively measure the importance of an individual dimension by quantifying its influence on the power of the MMD test. The KIM-based variable selection methods are model-free, work for high-dimensional data, and can capture important variables under different models. Theoretical properties of the KIM are provided and supported by simulation results.