Generalized maximum entropy estimation

We consider the problem of estimating a probability distribution that maximizes the entropy while satisfying a finite number of moment constraints, possibly corrupted by noise. Based on duality of convex programming, we present a novel approximation scheme using a smoothed fast gradient method that is equipped with explicit bounds on the approximation error. We further demonstrate how the presented scheme can be used for approximating the chemical master equation through the zero-information moment closure method, and for an approximate dynamic programming approach in the context of constrained Markov decision processes with uncountable state and action spaces.

À propos de ce résultat

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Concepts associés

Chargement

Publications associées

Chargement

John Lygeros, Tobias Sutter
2019
Article

Résumé

Official source

https://infoscience.epfl.ch/record/271818?ln=fr

À propos de ce résultat

Concepts associés (8)

Concepts associés

Chargement

Publications associées (2)

Publications associées

Chargement

From Infinite to Finite Programs: Explicit Error Bounds with Applications to Approximate Dynamic Programming

Daniel Kuhn, John Lygeros, Peyman Mohajerin Esfahani, Tobias Sutter

We consider linear programming (LP) problems in infinite dimensional spaces that are in general computationally intractable. Under suitable assumptions, we develop an approximation bridge from the infinite-dimensional LP to tractable finite convex programs in which the performance of the approximation is quantied explicitly. To this end, we adopt the recent developments in two areas of randomized optimization and first order methods, leading to a priori as well as a posteriori performance guarantees. We illustrate the generality and implications of our theoretical results in the special case of the long-run average cost and discounted cost optimal control problems for Markov decision processes on Borel spaces. The applicability of the theoretical results is demonstrated through a constrained linear quadratic optimal control problem and a fisheries management problem.

2018

Chargement

Decision rules for information discovery in multi-stage stochastic programming

Daniel Kuhn

Stochastic programming and robust optimization are disciplines concerned with optimal decision-making under uncertainty over time. Traditional models and solution algorithms have been tailored to problems where the order in which the uncertainties unfold is independent of the controller actions. Nevertheless, in numerous real-world decision problems, the time of information discovery can be influenced by the decision maker, and uncertainties only become observable following an (often costly) investment. Such problems can be formulated as mixed-binary multi-stage stochastic programs with decision-dependent non-anticipativity constraints. Unfortunately, these problems are severely computationally intractable. We propose an approximation scheme for multi-stage problems with decision-dependent information discovery which is based on techniques commonly used in modern robust optimization. In particular, we obtain a conservative approximation in the form of a mixed-binary linear program by restricting the spaces of measurable binary and real-valued decision rules to those that are representable as piecewise constant and linear functions of the uncertain parameters, respectively. We assess our approach on a problem of infrastructure and production planning in offshore oil fields from the literature.

IEEE2011

Chargement

Optimisation convexe

vignette|320x320px|Optimisation convexe dans un espace en deux dimensions dans un espace contraint E L'optimisation convexe est une sous-discipline de l'optimisation mathématique, dans la

Processus de décision markovien

En théorie de la décision et de la théorie des probabilités, un processus de décision markovien (en anglais Markov decision process, MDP) est un modèle stochastique où un agent prend des décisions et

Programmation dynamique

En informatique, la programmation dynamique est une méthode algorithmique pour résoudre des problèmes d'optimisation. Le concept a été introduit au début des années 1950 par Richard Bellman. À l'épo