A measure theoretical approach to the mean-field maximum principle for training NeurODEs
From MaRDI portal
Publication:2105521
DOI10.1016/j.na.2022.113161zbMath1503.49005arXiv2107.08707OpenAlexW4309287429MaRDI QIDQ2105521
Benoît Bonnet, Hui Huang, Massimo Fornasier, Cristina Cipriani
Publication date: 8 December 2022
Published in: Nonlinear Analysis. Theory, Methods \& Applications. Series A: Theory and Methods (Search for Journal in Brave)
Full work available at URL: https://arxiv.org/abs/2107.08707
Optimality conditions for problems involving randomness (49K45) Existence theories for optimal control problems involving relations other than differential equations (49J21)
Related Items
Optimal control of nonlocal continuity equations: numerical solution ⋮ Optimal control problems in transport dynamics with additive noise ⋮ Optimal control of ensembles of dynamical systems ⋮ Control of neural transport for normalising flows
Uses Software
Cites Work
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Constructive quantization: approximation by empirical measures
- On the rate of convergence in Wasserstein distance of the empirical measure
- Semiconcave functions, Hamilton-Jacobi equations, and optimal control
- Necessary optimality conditions for optimal control problems in Wasserstein spaces
- The phase flow method
- Forward-backward stochastic differential equations and controlled McKean-Vlasov dynamics
- A priori estimates for operational differential inclusions
- Applied functional analysis. Functional analysis, Sobolev spaces and elliptic differential equations
- Mean field games
- An introduction to \(\Gamma\)-convergence
- The Pontryagin Maximum Principle in the Wasserstein space
- Mean field control hierarchy
- Mean-field Pontryagin maximum principle
- Provable approximation properties for deep neural networks
- Applied functional analysis. Main principles and their applications
- Semiconcavity and sensitivity analysis in mean-field optimal control and applications
- DNN expression rate analysis of high-dimensional PDEs: application to option pricing
- Lagrangian, Eulerian and Kantorovich formulations of multi-agent optimal control problems: equivalence and gamma-convergence
- Optimal approximation of piecewise smooth functions using deep ReLU neural networks
- Function approximation by deep networks
- Differential inclusions in Wasserstein spaces: the Cauchy-Lipschitz framework
- Deep learning as optimal control problems: models and numerical methods
- A mean-field optimal control formulation of deep learning
- Sharp asymptotic and finite-sample rates of convergence of empirical measures in Wasserstein distance
- A proposal on machine learning via dynamical systems
- Deep vs. shallow networks: An approximation theory perspective
- Mean-Field Optimal Control
- Mean-field sparse optimal control
- A WELL-POSEDNESS THEORY IN MEASURES FOR SOME KINETIC MODELS OF COLLECTIVE MOTION
- Control in the Spaces of Ensembles of Points
- Elliptic Partial Differential Equations of Second Order
- Stable architectures for deep neural networks
- Intrinsic Lipschitz Regularity of Mean-Field Optimal Controls
- Neural ODEs as the deep limit of ResNets with constant weights
- Deep Neural Network Approximation Theory
- Spatially Inhomogeneous Evolutionary Games
- Analysis of the Generalization Error: Empirical Risk Minimization over Deep Artificial Neural Networks Overcomes the Curse of Dimensionality in the Numerical Approximation of Black--Scholes Partial Differential Equations
- A Pontryagin Maximum Principle in Wasserstein spaces for constrained optimal control problems
- Mean-field optimal control as Gamma-limit of finite agent controls
- Mean Field Games and Mean Field Type Control Theory
- Learning representations by back-propagating errors
- Understanding Machine Learning
- Mean-Field Optimal Control and Optimality Conditions in the Space of Probability Measures
