Differentiable Submodular Maximization

被引:0
作者
Tschiatschek, Sebastian [1 ]
Sahin, Aytunc [2 ]
Krause, Andreas [2 ]
机构
[1] Microsoft Res Cambridge, Cambridge, England
[2] Swiss Fed Inst Technol, Zurich, Switzerland
来源
PROCEEDINGS OF THE TWENTY-SEVENTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE | 2018年
基金
瑞士国家科学基金会; 欧洲研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We consider learning of submodular functions from data. These functions are important in machine learning and have a wide range of applications, e.g. data summarization, feature selection and active learning. Despite their combinatorial nature, submodular functions can be maximized approximately with strong theoretical guarantees in polynomial time. Typically, learning the submodular function and optimization of that function are treated separately, i.e. the function is first learned using a proxy objective and subsequently maximized. In contrast, we show how to perform learning and optimization jointly. By interpreting the output of greedy maximization algorithms as distributions over sequences of items and smoothening these distributions, we obtain a differentiable objective. In this way, we can differentiate through the maximization algorithms and optimize the model to work well with the optimization algorithm. We theoretically characterize the error made by our approach, yielding insights into the tradeoff of smoothness and accuracy. We demonstrate the effectiveness of our approach for jointly learning and optimizing on synthetic maximum cut data, and on real world applications such as product recommendation and image collection summarization.
引用
收藏
页码:2731 / 2738
页数:8
相关论文
共 50 条
[21]   Streaming adaptive submodular maximization* [J].
Tang, Shaojie ;
Yuan, Jing .
THEORETICAL COMPUTER SCIENCE, 2023, 944
[22]   On the Complexity of Dynamic Submodular Maximization [J].
Chen, Xi ;
Peng, Binghui .
PROCEEDINGS OF THE 54TH ANNUAL ACM SIGACT SYMPOSIUM ON THEORY OF COMPUTING (STOC '22), 2022, :1685-1698
[23]   The FAST Algorithm for Submodular Maximization [J].
Breuer, Adam ;
Balkanski, Eric ;
Singer, Yaron .
25TH AMERICAS CONFERENCE ON INFORMATION SYSTEMS (AMCIS 2019), 2019,
[24]   Distributionally Robust Submodular Maximization [J].
Staib, Matthew ;
Wilder, Bryan ;
Jegelka, Stefanie .
22ND INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS, VOL 89, 2019, 89 :506-516
[25]   Maximization of Approximately Submodular Functions [J].
Horel, Thibaut ;
Singer, Yaron .
ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 29 (NIPS 2016), 2016, 29
[26]   Practical Budgeted Submodular Maximization [J].
Feldman, Moran ;
Nutov, Zeev ;
Shoham, Elad .
ALGORITHMICA, 2023, 85 (05) :1332-1371
[27]   Regularized nonmonotone submodular maximization [J].
Lu, Cheng ;
Yang, Wenguo ;
Gao, Suixiang .
OPTIMIZATION, 2024, 73 (06) :1739-1765
[28]   Budgeted Sequence Submodular Maximization [J].
Chen, Xuefeng ;
Feng, Liang ;
Cao, Xin ;
Zeng, Yifeng ;
Hou, Yaqing .
PROCEEDINGS OF THE THIRTY-FIRST INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, IJCAI 2022, 2022, :4733-4739
[29]   ORACLE COMPLEXITY OF SUBMODULAR FUNCTION MAXIMIZATION [J].
KOVALEV, MM ;
MOSHCHENSKY, AV .
DOKLADY AKADEMII NAUK BELARUSI, 1992, 36 (02) :111-114
[30]   Robust monotone submodular function maximization [J].
Orlin, James B. ;
Schulz, Andreas S. ;
Udwani, Rajan .
MATHEMATICAL PROGRAMMING, 2018, 172 (1-2) :505-537