DIFFERENTIABLE PROGRAMMING A LA MOREAU

被引：0

作者：

Roulet, Vincent ^{[1
]}

Harchaoui, Zaid ^{[1
]}

机构：

[1] Univ Washington, Dept Stat, Seattle, WA 98195 USA

来源：

2022 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP) | 2022年

关键词：

Moreau envelope; differentiable programming; mathematical optimization; deep learning;

D O I：

10.1109/ICASSP43922.2022.9746423

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

The notion of a Moreau envelope is central to the analysis of first-order optimization algorithms for machine learning and signal processing. We define a compositional calculus adapted to Moreau envelopes and show how to apply it to deep networks, and, more broadly, to learning systems equipped with automatic differentiation and implemented in the spirit of differentiable programming.

引用

页码：3498 / 3502

页数：5

共 28 条

[1]

Abadi Martin, 2016, Proceedings of OSDI '16: 12th USENIX Symposium on Operating Systems Design and Implementation. OSDI '16, P265

[2]

Ahmad Nasir, 2020, ADV NEURAL INFORM PR, V33

[3]

[Anonymous], 2020, Advances in Neural Information Processing Systems

[4]

[Anonymous], P 17 INT C ART INT S

[5]

[Anonymous], 2015, JOINT EUR C MACH LEA, DOI DOI 10.1007/978-3-319-23528-831

[6]

Attouch Hedy, 1977, CR HEBD ACAD SCI, V284, P13

[7]

Bauschke H. H., 2017, Convex analysis and monotone operator theory in Hilbert spaces, V408

[8]

Bolte J., 2020, ADV NEUR IN, V33

[9]

Boski M, 2017, 2017 10TH INTERNATIONAL WORKSHOP ON MULTIDIMENSIONAL (ND) SYSTEMS (NDS)

[10] Efficiency of minimizing compositions of convex functions and smooth maps [J].

Drusvyatskiy, D. ;

Paquette, C. .

MATHEMATICAL PROGRAMMING, 2019, 178 (1-2) :503-558

← 1 2 3 →