Bilevel optimization for automated machine learning:a new perspective on framework and algorithm

作     者：Risheng Liu Zhouchen Lin 

作者机构：School of Software Technology Dalian University of Technology School of Intelligence Science and Technology Peking University 

出 版 物：《National Science Review》 (国家科学评论(英文版))

年 卷 期：2024年第11卷第8期

页      面：17-19页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported by the National Key R&D Program of China (2022YFA1004101) the National Natural Science Foundation of China(U22B2052 and 62276004) 

主　　题：automated machine learning bilevel optimization meta feature learning neural architecture search hyperparameter optimization 

摘      要：Machine learning (ML) has witnessed an unprecedented evolution in recent years,becoming a key driver of building artificial intelligence systems. With cuttingedge technologies such as Alpha Go [1]and ChatGPT [2], the power and versatility of ML have been demonstrated across diverse applications. However, designing effective ML solutions in realworld application scenarios can be challenging and time-consuming, thus paving the way for the emergence of automated machine learning (AutoML).