Big Data Research ›› 2021, Vol. 7 ›› Issue (5): 100-110.doi: 10.11959/j.issn.2096-0271.2021051

• COLUMN: DATA-DRIVEN OPTIMIZATION • Previous Articles     Next Articles

Optimization from samples

Zhijie ZHANG1,2, Xiaoming SUN1,2, Jialin ZHANG1,2, Wei CHEN3   

  1. 1 Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100086, China
    2 University of Chinese Academy of Sciences, Beijing 100049, China
    3 Microsoft Research Asia, Beijing 100080, China
  • Online:2021-09-15 Published:2021-09-01
  • Supported by:
    The National Natural Science Foundation of China(61832003);The National Natural Science Foundation of China(61872334);The Strategic Priority Research Program of Chinese Academy of Sciences(XDA27000000)

Abstract:

Optimization from samples studies how one can optimize objective functions from the sample data that one uses to learn them.Firstly, the mathematical model of this problem-optimization from samples model, as well as the inapproximability results under this model, was introduced.Secondly, some approaches and variants of OPS were introduced, in order to circumvent the impossibility results and make optimization possible.Thirdly, one of the variants-the optimization from structured samples model was focused on, and the algorithms for maximum coverage and influence maximization problem under it were introduced in details.Finally, the paper was concluded, and some future research directions for the problem were proposed.

Key words: optimization from samples, data-driven optimization, structured sample, maximum coverage problem, influence maximization problem

CLC Number: 

No Suggested Reading articles found!