14h00 | Départ des navettes depuis la gare de Grenoble | |||||
15h30-18h30 | mini-cours GdR MOA | |||||
19h30-21h30 | Dîner |
9h-12h | mini-cours GdR MOA | |||||
12h-14h | Repas | |||||
14h00-15h30 | Temps libre (travail ou ski) | |||||
15h30-18h30 | mini-cours GdR MOA | |||||
18h15 | Départ des navettes depuis la gare de Grenoble | |||||
17h-20h30 | Accueil des participants à la conférence | |||||
20h-22h | Dîner |
8h45-9h00 | Ouverture des journées | |||||
9h-10h | Luis NUNES VICENTE Local and Global Direct Search and an application to Additive Manufacturing (3D Printing) We have developed a new method for unconstrained and linearly constrained global optimization when derivatives of the objective function are unavailable or are difficult to obtain. The goal is to derive a fully parallelizable method for the efficient and accurate solution of large-scale problems. The methodology is based on multistarting probabilistic direct search, a technique for optimizing without derivatives in the context of local optimization. The initial multistarted set of runs can be split or merged based on the previously evaluated points, either according to their space clusterization (with no use of their function values) or according to the solution of appropriate nonconvex model subproblems (now using their function values). We provide numerical results on a large set of nonconvex unconstrained and bound-constrained problems. We will describe the application of our global solver to the optimization of object orientation in additive manufacturing (3D printing), in the context of an applied industrial project. Local and Global Direct Search and an application to Additive Manufacturing (3D Printing) Modérateur: Serge GRATTON, salle plénière (Vercors) | |||||
10h-10h30 | Pause Café | |||||
10h30-12h |
Dynamics and games (CA16228 GAMENET session)
Dynamics and games (CA16228 GAMENET session) Guilherme MAZANTI
Cet exposé s'intéresse à un modèle de jeu à champ moyen motivé par les mouvements de foule, où l'objectif de chaque agent est de sortir d'un domaine donné par une partie de son bord en temps minimal. Chaque agent est libre de se déplacer dans toutes les directions, mais sa vitesse maximale est bornée en termes de la distribution d'agents autour de sa position afin de prendre en compte des phénomènes de congestion.
Dans un premier temps, nous formulerons ce jeu à champ moyen dans un cadre Lagrangien et montrerons l'existence d'un équilibre par une méthode de point fixe. Grâce à des propriétés supplémentaires de régularité des trajectoires optimales des agents obtenues par le principe du maximum de Pontryagin, nous caractériserons ensuite les équilibres à travers un système d'une équation de continuité sur la distribution d'agents couplée avec une équation de Hamilton--Jacobi sur la fonction valeur du problème de contrôle optimal de chaque agent. Nous montrerons finalement que, si la distribution des agents à l'instant initial est une mesure absolument continue à densité $L^p$, alors cette mesure reste à densité $L^p$ pour tout instant $t \geq 0$. Jeux à champ moyen en temps minimal
Xavier VENEL, Bruno ZILIOTTO
In a Markov Decision Process (MDP), at each stage, knowing the current state, the decision-maker chooses an action, and receives a reward depending on the current state of the world. Then a new state is randomly drawn from a distribution depending on the action and on the past state. Many optimal payoffs concepts have been introduced to analyze the strategic aspects of MDPs with long duration: asymptotic value, uniform value, liminf average payoff criterion, limsup average payoff criterion… We provide sufficient conditions under which these concepts coincide completing previous results of Renault and Venel (2016) and Venel and Zilioto (2016). Markov Decision Processes with long duration
|
Composite Optimization
Composite Optimization Guillaume GARRIGOS, Lorenzo ROSASCO, Silvia VILLA
We consider inverse problems in separable Hilbert spaces where the prior on the data is an assumption of structured sparsity.
We look at a class of regularizers for which
minimization algorithms identify in finite time the extended support of the original data.
This is a direct consequence of a more general identification theorem, involving the mirror stratifiability of the regularizer, a notion developped recently, and based on duality arguments.
As a by-product, we derive improved rates of convergence for the minimization algorithms, like a new linear rate result for the soft-thresholding algorithm in $\ell^2(\mathbb{N})$ with no assumptions.
We then provide necessary and sufficient conditions for norm regularizers to be mirror stratifiable, and show its tight relationship
with the geometry of the corresponding unit ball.
We apply this characterization result to show that norm regularizers inducing group sparsity with overlap are not mirror-stratifiable.
We then discuss how to adapt the notion of mirror-stratifiability to treat these regularizers. Structured sparsity in inverse problems and support recovery with mirror-stratifiable functions
Philippe MAHEY, Ernesto ORE ALBORNOZ, Eladio OCANA
Operator splitting methods have been recently concerned with inclusions problems based on composite operators made of the sum of two monotone operators, one of them associated with a linear transformation. We analyze here a general and new splitting method which indeed splits both operator proximal steps, and avoids costly numerical algebra on the linear operator. The family of algorithms induced by our generalized setting includes known methods like Chambolle-Pock primal-dual algorithm and Shefi-Teboulle’s Proximal Alternate Direction Method of Multipliers.
A convergence analysis is given and separable augmented Lagrangian algorithms are derived from our scheme to solve convex optimization problems with separable structure and coupling constraints. A unified splitting method for composite monotone inclusions
Olivier FERCOQ, Zheng QU
Nous proposons de nouvelles stratégies de redémarrage pour la méthode de descente par coordonnée accélérée. Notre contribution principale est de montrer que pour une suite bien choisie d'instants de redémarrage, la méthode redémarrée a un taux de convergence presque géométrique. Une caractéristique majeure de la méthode est qu'elle profite de la borne d'erreur quadratique locale sans utiliser explicitement sa valeur effective. Nous montrons aussi que sous l'hypothèse plus restrictive de forte convexité, un redémarrage à période fixe donne une taux de convergence géométrique, quelle que soit la période. Nous illustrons les performances de l'algorithme sur un problème de régression logistique et sur un problème de Lasso.
Redémarrer la méthode de descente par coordonnée accélérée avec une estimation imprécise de la borne d'erreur quadratique
| ||||
12h-14h | Repas | |||||
14h-15h30 |
Control Theory
Control Theory Térence BAYEN, Olivier COTS
We consider a minimal time control problem governed by an affine system w.r.t. the control. Our aim is to study properties of the optimal synthesis in presence of a singular locus that involves a saturation point for the singular control. We show that the optimal synthesis exhibits a prior saturation point at the intersection of the singular arc and a switching curve,
and we also provide qualitative properties of this curve. We highlight this phenomenon on several models arising in nuclear magnetic resonance and in bioprocesses. About prior saturation points for affine control systems
Ivan YEGOROV, Francis MAIRET, Hidde DE JONG, Jean-Luc GOUZE
Microorganisms have evolved complex strategies for controlling the distribution of available resources over cellular functions. Biotechnology aims at interfering with these strategies, so as to optimize the production of metabolites and other compounds of interest, by (re)engineering the underlying regulatory networks of the cell. The resulting reallocation of resources can be described by simple so-called self-replicator models and the maximization of the synthesis of a product of interest formulated as a dynamic optimal control problem. Motivated by recent experimental work, we are specifically interested in the maximization of metabolite production in cases where growth can be switched off through an external control signal. We study various optimal control problems for the corresponding self-replicator models by means of a combination of analytical and computational techniques. We show that the optimal solutions for biomass maximization and product maximization are very similar in the case of unlimited nutrient supply, but diverge when nutrient are limited. Moreover, external growth control overrides natural feedback growth control and leads to an optimal scheme consisting of a first phase of growth maximization followed by a second phase of product maximization. This two-phase scheme agrees with strategies that have been proposed in metabolic engineering. More generally, our work shows the potential of optimal control theory for better understanding and improving biotechnological production processes. Optimal control of bacterial growth for the maximization of metabolite production
Denis ARZELIER, Florent BREHARD, Mioara JOLDES, Jean-Bernard LASSERRE, Aude RONDEPIERRE
Dans cet exposé, nous proposons une nouvelle modélisation pour le calcul de la probabilité de collision entre deux objets sphériques en orbite autour de la Terre. Dans un premier temps, nous montrerons comment, à l'aide des mesures d'occupation, le calcul de la probabilité de collision peut se reformuler en un problème de programmation linéaire de dimension infinie dans le cône des mesures semi-définies positives (SDP), dont la valeur optimale est exactement la probabilité cherchée. Dans un second temps, nous verrons comment résoudre numériquement ce problème d'optimisation linéaire dans l'espace des mesures par la manipulation de leurs moments en construisant une hiérarchie de relaxations SDP. On construit ainsi une suite de majorants convergeant vers la probabilité de collision cherchée. Nous conclurons cet exposé par quelques tests numériques montrant les forces et les limitations de cette approche au regard de l'application visée. Probabilité globale de collision - Une modélisation par les mesures d'occupation
|
Stochastic Optimization
Stochastic Optimization Vincent LECLERE
The Stochastic Dual Dynamic Programming (SDDP) algorithm
has become one of the main tools to address
convex multistage stochastic optimal control problem.
Recently a large amount of work has been devoted to improve
the convergence speed of the algorithm through cut-selection and regularization,
or to extend the field of applications to non-linear, integer or risk-averse problems.
However one of the main downside of the algorithm remains the difficulty to
give an upper bound of the optimal value, usually estimated through Monte Carlo
methods and therefore difficult to use in the algorithm stopping criterion.
In this paper we present a dual SDDP algorithm that yields a converging
exact upper bound for the optimal value of the optimization problem.
Incidently we show how to compute an alternative control policy based
on an inner approximation of Bellman value functions instead of
the outer approximation given by the standard SDDP algorithm.
We illustrate the approach on an energy production problem involving
zones of production and transportation links between the zones.
The numerical experiments we carry out on this example show
the effectiveness of the method. Exact converging bounds for Stochastic Dual Dynamic Programming
Welington DE OLIVEIRA, Wim VAN ACKOOIJ, Yongjia SONG
We consider well-known decomposition techniques for multistage stochastic programming and a new scheme based on normal solutions for stabilizing iterates during the solution process. The given algorithms combine ideas from finite perturbation of convex programs and level bundle methods to regularize the so-called forward step of these decomposition methods. Numerical experiments on a hydrothermal scheduling problem indicate that our algorithms are competitive with the state-of-the-art approaches such as multistage regularized decomposition and stochastic dual dynamic programming.
On level regularization with normal solutions in decomposition methods for multistage stochastic programming problems
Sofia MICHEL
We propose a data-driven method to optimize public transport schedules. Using transit data to construct scenarios that reflect the uncertainty of the system, we compute schedules that minimize the expected waiting time during transfers. We model the problem as a two-stage stochastic program and propose two equivalent formulations: a mixed integer linear program (MILP) and a mixed integer quadratic program (MIQP). The MILP version is solved exactly using a generic solver but does not scale well; whereas the MIQP has a partially separable structure that we exploit to design an efficient local search heuristic. We provide results of the two approaches and compare them to the state of the art using transit data collected from Nancy, France. Stochastic Optimization of Public Transport Schedules
| ||||
15h30-16h00 | Pause Café | |||||
16h-17h | Guillaume OBOZINSKI Un lagrangien dual et un schéma par blocs stochastique pour l'apprentissage de modèles de champs de Markov conditionnels L'apprentissage de modèles de champs de Markov est traditionnellement très coûteuse car elle requiert classiquement de résoudre à chaque itération un problème d'inférence probabiliste sur le graphe associé. Je présenterai dans cet exposé une formulation lagrangienne augmentée duale pour l'apprentissage de modèles de champs de Markov conditionnels. Celle-ci permet la construction d'un algorithme qui peut être interprété comme une descente de gradient inexact sur le multiplicateur de Lagrange et ne nécessite pas de résoudre le problème d'inférence probabiliste dans le modèle à chaque itération, mais requiert seulement d'effectuer un nombre fixe d'itérations proximales par bloc randomisées sur les cliques du graphe. L'algorithme est globalement linéairement convergent en terme du nombre total de mise à jour des paramètres duaux associés aux cliques, et montre de très bonnes performances en pratique. (En collaboration avec Shell Xu Hu) Un lagrangien dual et un schéma par blocs stochastique pour l'apprentissage de modèles de champs de Markov conditionnels Modérateur: Julien MARIAL, salle plénière (Vercors) | |||||
17h-18h | Spotlight posters | |||||
18h-19h30 |
Session Poster / Apéritif
Session Poster / Apéritif Amélie HELIOU, Johanne COHEN, Panayotis MERTIKOPOULOS
Motivés par des défis actuels (réseaux, biologie, ...) nous avons étudié des algorithmes qui peuvent être appliqués à un grand nombre de joueurs qui ont une connaissance limitée du jeu.
Dans de tels jeux, les algorithmes de non regret sont largement utilisés.
Ces algorithmes garantissent que la différence entre les sommes dans le temps des revenus d'un joueur et des revenus de sa meilleure stratégie est sous-linéaire.
Les équilibres de Nash sont des états où aucun joueur ne bénéficierait de changer seul de stratégie.
Des études récentes ont montré que la limite de la séquence de stratégies de certains types d'algorithmes de non regret peut être arbitrairement proche d'un équilibre de Nash avec une probabilité proche de $1$.
Les équilibres de Nash peuvent-ils être la limite presque surement d'un algorithme de non regret ?
Nous répondons positivement à cette question en analysant l'algorithme Hedge qui a la propriété de non regret.
Nous nous plaçons dans le cadre d'une information imparfaite (bandit), où les joueurs ont uniquement accès à une estimation du revenu rapporté par la stratégie pure qu'ils ont joué.
Nous montrons que lorsque Hedge est appliqué à des jeux de potentiel génériques, la séquence de stratégies obtenues converge vers un équilibre de Nash. Apprentissage dans le cas bandit sur des jeux de potentiel
Samer DWEIK
Dans cet exposé, nous considérons le problème de minimisation de la variation totale, en 2D, suivant
$$ \min\{\int_{\Omega} |\nabla u| dx : u \in BV(\Omega), u=g on \partiel\Omega\}$$ (GP)
où $g: \partial\Omega \mapsto \mathbb{R}$ est une fonction $L^1$ donnée et $\Omega \subset \mathbb{R}^2$ est un domaine uniformément convexe.
Tout d'abord, on montrera comment ce problème est, en fait, équivalent au problème de Beckmann suivant
$$ \min\{\int_{\Omega} |w| dx : w \in L^1(\Omega,\mathbb{R}^2), \div(w)=0, w.n=f on \partial\Omega\}$$ (BP)
où $f$ est la dérivée tangentielle de $g$.
Ainsi, nous abordons le problème de la régularité $W^{1, p}$ pour une solution $u$ de (GP) en étudiant la sommabilité $L^p$
du flot minimal $w$ de (BP) quand on transporte des mesures concentrées sur le bord. Plus précisément, nous montrons que pour une donnée au bord $g$ dans $W^{1, p}$, la solution $u$ est aussi dans $W^{1, p}$, dès que $p \leq 2$. D'autre part, par un contre-exemple, nous montrons que ce résultat échoue pour $p>2$. Une approche par transport optimal à la minimisation de la variation totale en 2D
Boubacar FALL, Diaraf SECK, Filippo SANTAMBROGIO
En considérant un domaine donné $R$ contenant un obstacle $O$, nous allons établir la dérivée première de forme du temps moyen de sortie des partilucles quand ces dernières sortent du domaine suivant une EDP modélisant le mouvement de foule. Cette fonction qui est le temps moyen de sortie dépend effectivement de la géométrie de l'obstacle $O$. Cette dérivée première est un pas important pour la minimisation du temps moyen de sortie des particles dans le domaine $R$. Dans ce travail, nous allons considérer la famille suivante d'équations aux dérivées partielles: Fokker-Planck, le cas des Mileux poreux et un cas plus général. En outre, dans le cas de dimension deux nous allons expliciter l'expression de la dérivée de forme mettant en evidence le théoreme de Hadamard. Formules de dérivées de forme pour les obstacles en mouvements de foules
Dong Quan VU, Patrick LOISEAU, Alonso SILVA
The Colonel Blotto game is a well-known strategic resource allocation problem with important applications in a vast range of domains, from advertising to security. Two players simultaneously allocate a fixed budget of resources across $n$ battlefields, each player trying to maximize the aggregate value of the battlefields where she has higher allocation. Despite its long-standing history and importance, the Colonel Blotto game still lacks a complete Nash equilibrium solution in its general form with asymmetric players' budgets and heterogeneous battlefield values.
In this paper, we propose an approximate equilibrium for this general case. We construct a simple strategy (the independently uniform strategy) and prove that it is an epsilon-equilibrium. We give a theoretical bound on the approximation error in terms of the number of battlefields and players’ budgets which identifies precisely the parameters regime for which our strategy is a good approximation. We also investigate an extension to the discrete version (where players can only have integer allocations), for which we proposed a modified strategy that can be computed very efficiently and gives an epsilon-equilibrium of the game. Approximate equilibria of Blotto-type games
Stéphane DURAND, Federica GARIN, Bruno GAUJAL
In this paper we design and analyze a distributed algorithm to
compute a Nash equilibrium in potential games based on best response.
This algorithm can be implemented in a distributed way using Poisson
clocks and does not rely on the usual assumption that players take no
time to compute their best response and change their strategy. We show
that if one takes this time lag into account, the algorithm may suffer from
collisions (one player starts to play while another player has not changed her
strategy yet). We first show that collisions do not jeopardize convergence
to a Nash equilibrium but our main result is to show that the average
time complexity of the algorithm (time to reach a Nash equilibrium) is
bounded by 2n log n + o(n log n), where n is the number of players and p is 1−p
the probability that a player gets in a collision. Furthermore, exhaustive simulations suggest that the term log n is superfluous. Distributed Best Response in Random Potential Games
Antoine RICHARD, Brice MAYAG, Yves MEINARD, François TALBOT, Christophe RIOU, Alexis TSOUKIAS
Physicians at the Hospices Civils of Lyon (HCL) have at their disposal a Health Information System (HIS) which provides them access to all the available information about their patients. Our goal is to identify relevant improvements that we can make on the current HIS used by physicians at the HCL, to help them during the diagnostic process. According to this aim, we develop an analysis and a model of clinical diagnostic process. We first show that the decision process unfolded by a physician is mainly based on searching information about the patient. Then, we propose a rule-based model of the decision process to detect which information the physician will need and when s/he will need it during her/his diagnostic process. An analysis of the decision process of a clinical diagnostic: from empirical observation to rule-based modelling
Wael SAKER, Philippe BICH
Many papers have recently focused on the issue of the existence of a Nash equilibrium of {\it normal-form} games with {\it discontinuous payoffs}. For {\it extensive-form } games, several papers have provided existence proof of a subgame perfect equilibrium (in short, SPE, a generalization of Nash equilibrium) for infinite sets of actions at each period, when the payoff functions are assumed to be {\it continuous}.
In this paper, we provide the first general existence result of a SPE for a class of discontinuous extensive-form games. This class is called extensive-form better-reply secure, and has some similarity, in terms of definition, with the class of better-reply secure games.\footnote{The class of better-reply secure games was introduced by Reny in 1999. It is a class of discontinuous normal-form games for which a Nash equilibrium can be proved to exist.} Roughly, a game is extensive-form better-reply secure if for every strategy profile $s^*$ which is not a subgame perfect equilibrium, there is an history $h$ from time 0 to time i (where player $i$ has to play), such that for every payoff $u_i$ of player $i$ obtained as a limit of the payoffs generated by some paths begining at $h$, there exists a deviation of player $i$ beginning at $h$, which gives him a payoff strictly above $u_i^*$ , even if one perturbs slightly the path defined by $s^*$ after $h$. Our result is the counterpart of Reny's existence theorem (true only for normal form games) for extensive form games. The idea is to approximate the game not by {\it a given finite approximation} (which would raise the question of its choice), as it is usually done, but by {\it any finite approximation}. This allows to improve the quality of the existence result, and relax the assumption of continuity of payoffs. We provide several examples of extensive-form better-reply secure games. On The Existence Of Subgame Perfect Equilibria In Discontinuous Perfect Information Games
Tristan GARREC
Nous présentons deux jeux à deux joueurs et à somme nulle modélisant une situation dans laquelle l'un d'eux attaque (ou se cache) dans un ensemble compact en dimension finie, et l'autre tente d'empêcher l'attaque (ou de le trouver). Le premier jeu, dit jeu de patrouille, correspond à une formulation dynamique de cette situation, au sens où l'attaquant choisit un temps et un point d'attaque tandis que le patrouilleur choisit une trajectoire continue afin de maximiser la probabilité de trouver le point d'attaque dans un délais donné. Le second jeu, dit jeu de dissimulation, correspond à une formulation statique dans laquelle les deux joueurs choisissent simultanément un point, le chercheur maximise la probabilité d'être à une distance inférieure à un rayon donné de son adversaire. Jeux de patrouille et de dissimulation continus
Paulin JACQUOT
Routing games find many applications in various fields such as transportation, telecommunications or energy. The set of Nash Equilibria in this class of games is in general hard to compute. Here, we focus on the particular case of a parallel network and we address the computational issue of equilibria by providing two algorithms: the cycling best response dynamics and a projected gradient descent method. Under some monotonicity assumptions, we prove the convergence of those methods and we provide an upper bound on their convergence rate. Our convergence results state that, using one of these algorithms, the unique equilibrium of the game can be computed at an arbitrary precision in polynomial time. We give a practical application in the energy sector, where this framework and the associated results can be used to optimize the electricity consumption of flexible users Fast Computation of Equilibria in Monotone Routing Games and Application to Electricity Demand Response
Eugene NDIAYE, Tam LE, Olivier FERCOQ, Joseph SALMON, Ichiro TAKEUCHI
Various machine learning problem can be written as a minimization of a loss function $f$ plus a regularization $R$ calibrated by a positive hyperparameter $\lambda$ \ie $\min_{\beta} f(X\beta) + \lambda R(\beta)$. A major weakness of these methods is the tuning of the regularization parameter $\lambda$ where the difficulty are both theoretical (for statistical guarantees) as well as algorithmic (for numerical stability). A common approach is to choose the best parameter among the entire solution path on a given range. Unfortunately, these methods have a worst case complexity that is exponential in the number of parameter (Gartner et al 2012). In order to avoid this issue, approximation of the solution path was proposed and an optimal complexity was known to be ($1/\epsilon$) (Giesen et al 2010). Surprisingly, (Mairal and Yu 2012) show an important improvement to ($1/\sqrt{\epsilon}$) for the special case of the Lasso. Their result was generalized in (Giesen et al 2012) with a lower and upper bound of ($1/\sqrt{\epsilon}$) and derive a generic algorithm by assuming a polynomial lower bound on the objective function. Following those ideas, (Shibagaki et al 2015) have proposed the approximation of the validation path which is more natural for selecting a hyperparameter that are guaranteed to achieve a validation error close to the optimal one.
We revisit the approximation and validation path theory in a unified framework under general regularity assumptions that are commonly verified in machine learning problems. We apply it to examples that encompass classification and regression problem and provide a complexity analysis along with optimality guarantees. We highlight the relationship between the complexity of the approximation path and the regularity of the loss function and show that the previously proposed complexity analysis can be too pessimistic or even insufficient in some situations. Optimal approximation for regularization and validation path
Daniel KADNIKOV
We study the concept of information. Regarding a stochastic control problems model by Witsenhausen we look at a usual game theoretical problem without knowing a priori the order in which decisions are made by agents. This approach yields configurations that are not covered by classical theory. We are as well studying possible systems and classify them using four types of orderings. We show that the extensive form model is embedded into Witsenhausen model with the potential of embedding Bayesian games as well. Games with information. Witsenhausen intrinsic model
Paul ARMAND, Ngoc Nguyen TRAN
We present a primal-dual algorithm for solving a bound constrained
optimization problem. The algorithm uses a regularization technique
to handle the case where the second order sufficient optimality conditions
do not hold at a local minimum. The local convergence analysis is
done under a weaker assumption and is related to a local error bound condition. It is proved that locally the algorithm is superlinearly convergent.
Some examples are also given to see the advantage of this new algorithm. Local convergence property of a regularized primal-dual method for bound constrained optimization
Simon BOULMIER
The mixed-integer quadratically-constrained quadratic programs
are a very general class of instances that allow to model many reallife
problems. We will present improvements on convex relaxations of such
problems, and an augmented lagrangian dual approach to solve these relaxations.
The framework is then integrated in a non-linear branch-and-bound
algorithm to find near optimal bounds. An efficient approach for solving large-scale non-convex quadratic problems
Michel GRABISCH, Alexis POINDRON, Agnieszka RUSINOWSKA
We study a stochastic model of anonymous influence with conformist and anti-conformist individuals. Each agent with a "yes" or "no" initial opinion on a certain issue can change his opinion due to social influence. We consider anonymous influence, which depends on the number of agents having a certain opinion, but not on their identity. An individual is conformist/anti-conformist if his probability of saying "yes" increases/decreases with the number of "yes"- agents.
The seminal work of DeGroot (1974) and some of its extensions consider a non-anonymous influence in which agents update their opinions by using a weighted average of the opinions of their neighbors. We are interested in anonymous influence, which depends only on the number of individuals having a certain opinion and is not dependent on agents' identities. Forster and al (2013) investigate such kind of social influence by using the ordered weighted averages (commonly called OWA operators, Yager (1988) which are the unique anonymous aggregation functions. The authors departure from a general framework of influence based on aggregation functions Grabisch and Rusinowska (2013), where every individual updates his opinion by aggregating the agents' opinions which determines the probability that his opinion will be "yes" in the next period. Both frameworks of Forster and al (2013) and Grabisch and Rusinowska (2013) cover only positive influence (imitation), since by definition aggregation functions are nondecreasing, and hence cannot model anti-conformism.
In order therefore to extend the analysis of anonymous influence to anti-conformism, we assume that every agent has a coefficient of conformism which is a real number in [-1,1], with negative/positive values corresponding to anti-conformists/conformists. The two extreme values -1 and 1 represent a pure anti-conformist and a pure conformist, respectively, and the remaining values -- so called "mixed" agents. We consider two kinds of a society: without mixed agents, and with mixed agents who play randomly either as conformists or anti-conformists.
For both cases of the model, we deliver a qualitative analysis of convergence, i.e., find all absorbing classes and conditions for their occurrence. We find nineteen terminal classes, whose dynamics can be distinguished into three categories : terminal states, periodic classes and aperiodic classes.
Though we do not examine issues like speed of convergence, average time between peaks in aperiodic classes and other quantitative results, we emphasize the need for such problem to be examined later. Another case for the need of such future research is that simulations show that, from a qualitative point of view, a society represented by a given absorbing class may exhibit a behavior similar to another terminal class, sometimes temporarily; i.e the behavior between two terminal classes, though qualitatively different, can be undistinguishable on data.
We argue that anti-conformism can be a word labelling different contexts, such as incomplete information or bounded rationality, that is, a wide range of processes based on imitation, signaling or coordination. Based on this remark and simulations of time series, we suggest that if some econometric method could be invented as a tool to disentangle a conformist and anti-conformist behaviors from time series, then such model could be useful to model irregular periodicity patterns, amplifying oscillations, booms and bursts; in particular, if an econometric method was to be designed to recover parameters of our model from times series, it could probably be used to predict some of the shocks due to imitation processes. A model of anonymous influence with anti-conformist agents
Benoît TRAN, Marianne AKIAN, Jean-Philippe CHANCELIER
Nous étudions un problème de contrôle optimal discret en horizon fini, de dynamique linéaire, de coûts instantanés et final quadratiques et de contrôles à la fois continus et discrets.
Nous développons un algorithme et prouvons sa convergence en s'inspirant du travail effectué dans~\cite{Zheng} et~\cite{Zhengbis}.
Les opérateurs de Bellman associés à l'équation de la programmation dynamique de notre problème sont min-plus linéaires sur les formes quadratiques, permettant ainsi de déterminer explicitement les fonctions valeurs comme des infimums finis de formes quadratiques. Une telle approche n'est pas soumise à la fameuse malédiction de la dimension, toutefois le nombre de formes quadratiques nécessaires au
calcul de l'infimum croît exponentiellement avec le temps.
L'algorithme proposé ici sélectionne certaines formes quadratiques impliquées dans le calcul de l'infimum en testant si une forme quadratique améliore l'approximation courante en un point tiré aléatoirement uniformément sur la sphère. Dans notre cadre, le calcul explicite des opérateurs de Bellman à contrôle discret fixé est donné par une équation de Riccati et nous prouvons la convergence de notre algorithme en nous basant sur l'existence d'un ensemble borné, pour l'ordre de Loewner sur les matrices symétriques, stable par tous les opérateurs de Bellman. Un algorithme stochastique de type min-plus pour un problème de contrôle optimal déterministe
Alberto BIETTI, Julien MAIRAL
Stochastic optimization algorithms with variance reduction have proven successful for minimizing large finite sums of functions. Unfortunately, these techniques are unable to deal with stochastic perturbations of input data, induced for example by data augmentation. In such cases, the objective is no longer a finite sum, and the main candidate for optimization is the stochastic gradient descent method (SGD). We introduce a variance reduction approach for these settings when the objective is composite and strongly convex. The convergence rate outperforms SGD with a typically much smaller constant factor, which depends on the variance of gradient estimates only due to perturbations on a single example. Stochastic Optimization with Variance Reduction for Infinite Datasets with Finite Sum Structure
Kimon ANTONAKOPOULOS, Panayotis MERTIKOPOULOS
Motivated by Nesterov’s dual averaging method for solving (stochastic) convex programs and monotone variational inequalities, we propose a primal-dual algorithm for finding zeros of random, non-monotone operators. Given the close connection between the zero set of a maximal monotone operator and the set of solutions of a (Minty-type) variational inequality, we focus on a class of non-monotone operators for which the associated Minty variational inequality admits a solution. This property, which we call variational coherence, is wide enough to properly include all maximal monotone, pseudomonotone (“+” or “∗”), and other relevant classes of operators. Under mild assumptions for the randomness of the problem at hand (bounded second moments), we show that the algorithm converges to a zero point with probability 1, and we estimate its rate of convergence. A primal-dual algorithm for finding zeros of random non-monotone operators in Hilbert spaces.
Arthur MENSCH, Julien MAIRAL, Bertrand THIRION, Gaël VAROQUAUX
We present a matrix-factorization algorithm that scales to input matrices with both huge number of rows and columns. Learned factors may be sparse or dense and/or non-negative, which makes our algorithm suitable for dictionary learning, sparse component analysis, and non-negative matrix factorization. Our algorithm streams matrix columns while subsampling them to iteratively learn the matrix factors. At each iteration, the row dimension of a new sample is reduced by subsampling, resulting in lower time complexity compared to a simple streaming algorithm. Our method comes with convergence guarantees to reach a stationary point of the matrix-factorization problem. We demonstrate its efficiency on massive functional Magnetic Resonance Imaging data (2 TB), and on patches extracted from hyperspectral images (103 GB). For both problems, which involve different penalties on rows and columns, we obtain significant speed-ups compared to state-of-the-art algorithms. Stochastic Subsampling for Factorizing Huge Matrices
Nicolas LEBBE, Charles DAPOGNY, Karim HASSAN, Edouard OUDET, Alain GLIERE
Nous étudions l'optimisation de la forme de dispositifs nano-photoniques par la méthode des dérivées de forme d'Hadamard.
Nos recherches nous ont porté vers l'optimisation robuste (au pire cas) de ces dispositifs lorsque des incertitudes sont attendues soit au niveau de la longueur d'onde de la lumière incidente soit au niveau de la forme réalisée.
Mathématiquement on cherche à résoudre le problème
$$\max_{\Omega} \inf_{\|\delta\| \leq m} \mathcal{J}(\Omega,\delta)$$
où $\Omega$ est la forme recherchée, $\delta$ un paramètre représentant l'incertitude d'amplitude maximale $m$ et $\mathcal{J}$ un objectif, tel que la puissance en sortie, qui dépend de la forme par l'intermédiaire du champs électrique, solution des équations de Maxwell à l'intérieur du dispositif.
La plupart des méthodes proposées pour ce type de problème d'optimisation robuste repose sur l'ajout d'un terme de pénalisation au niveau de l'objectif.
Dans cet exposé nous présenterons les résultats que nous avons obtenus en appliquant une méthode d'optimisation multi-objectifs. Optimisation robuste de forme pour la nano-photonique
Saeed HADIKHANLOO, Francisco SILVA
In this presentation, we consider a finite version of Mean Field Games (MFGs) introduced by Gomes et al. (2010) and study, first the convergence of the fictitious play procedure to equilibrium and second, the relation of some finite MFGs with the continuous first order MFGs system. Learning and convergence analysis in finite mean field games
Dmitry GRISHCHENKO, Franck IUTZELER, Jerome MALICK, Massih-Reza AMINI
We propose an efficient distributed algorithm for solving regularized learning problems. In a distributed framework with a master machine coordinating the computations of many slave machines, our proximal-gradient algorithm allows local computations and sparse communications from slaves to master. Furthermore, with the $\ell_1$-regularizer, our approach automatically identifies the support of the solution, leading to sparse communications from master to slaves, with near-optimal support. We thus obtain an algorithm with two-way sparse communications. Distributed Optimization with Sparse Communications and Structure Identification
Benjamin DUBOIS, Jean-François DELMAS, Guillaume OBOZINSKI
In this work, we consider a descent algorithm for the Sparse Reduced-Rank Regression problem. A recent litterature revisited such non- convex problems based on an explicit parametrization of the low-rank matrix. However, no general convergence rate result was provided, in particular not in the nondifferentiable case. Reformulating the minimization problem and analyzing its geometry in a neighborhood of the optimal set, we show the Polyak-Łojasiewicz inequality or its extension to the nondifferentiable case are satisfied. Consequently, we establish the linear convergence of the proximal block-coordinate gradient algorithm in this neighborhood. A Fast Algorithm for Sparse Reduced-Rank Regression
Manh-Khang DAO
We consider an optimal control on networks in the spirit of the works of Achdou et al. (2013) and Imbert et al. (2013). The main new feature is that there are entry costs at the edges of the network leading to a possible discontinuous value function. We characterize the value function as the unique viscosity solution of a new Hamilton-Jacobi system. The uniqueness is a consequence of a comparison principle for which we give two different proofs, one with arguments from the theory of optimal control inspired by Achdou et al. (2014) and one based on partial differential equations techniques inspired by a recent work of Lions and Souganidis (2016). Hamilton-Jacobi equations for optimal control on networks with entry costs
Matteo TACCHI, Tillmann WEISSER, Jean-Bernard LASSERRE, Didier HENRION
L'article [HEN09] propose une méthode pour le calcul du volume d'un compact semi-algébrique basique, utilisant les hiérarchies moments / SOS développées dans [LAS09]. L'article [HEN14] s'inspire de ces travaux pour estimer la région d'attraction (R.O.A.) d'un compact semi-algébrique basique pour une E.D.O. polynomiale. Dans ces deux approches, le recours au calcul SDP pose un problème de passage à l'échelle : la complexité des algorithmes est polynomiale en la dimension du problème. Nous proposons ici une méthode originale et non triviale pour le calcul de volumes, transposable à l'estimation de R.O.A., qui permet de traiter des problèmes de dimension relativement élevée à condition qu'ils présentent une certaine structure creuse. Calcul Parcimonieux pour les Volumes
| |||||
19h30-21h30 | Dîner |
9h-10h | Wim VAN ACKOOIJ Probability constraints, convexity and applications in energy A probability function results from a measure acting on a random inequality system depending both on a decision and random vector. Probability constraints requests that such a system holds with high enough probability in order to ensure safety of the decision. Probability functions arise in many applications in engineering in general, and in energy management in particular, in so-called probability or chance-constraints. Then, generally speaking, a desirable property from the viewpoint of optimization is convexity. This should however be understood as convexity of the feasible set. In this talk, we will review the general framework of probability constraint and discuss several recent insights in deriving convexity statements for feasible sets to probability constraints, that take various forms depending on the underlying structure of the probability function itself. These results differ most notably from the classic convexity results in so much as that they assert convexity of only some upper level sets of the probability function. Probability constraints, convexity and applications in energy Modérateur: Sandrine CHAROUSSET, salle plénière (Vercors) | |||||
10h-10h30 | Pause Café | |||||
10h30-12h |
Optimal transport and image processing
Optimal transport and image processing Aude GENEVAY
Generative models are a class of statistical models that can easily be sampled from but whose likelihood is intractable because they are supported on low-dimensional manifolds in much higher-dimensional spaces. Inference for these models thus can’t be performed with standard methods using maximum likelihood. That’s why the ability to compare two degenerate probability distributions is a crucial factor in the estimation of these generative models. It is therefore no surprise that optimal transport (OT) metrics and their ability to handle measures with non-overlapping supports have emerged as a promising tool.
Yet, training generative models using OT raises formidable computational and statistical challenges, because of (i) the computational burden of eval- uating OT losses, (ii) their instability and lack of smoothness, (iii) the difficulty to estimate them, as well as their gradients, in high dimension. By introducing an OT-based loss called Sinkhorn loss, we can tackle these three issues by relying on two key ideas: (a) entropic smoothing, which turns the original OT loss into a differentiable and more robust quantity that can be computed using Sinkhorn fixed point iterations; (b) algorithmic (automatic) differentiation of these iterations with seamless GPU execution. Additionally, Entropic smoothing generates a family of losses interpolating between Wasserstein (OT) and Energy distance/Maximum Mean Discrepancy (MMD) losses, thus allowing to find a sweet spot leveraging the geometry of OT on the one hand, and the favorable high-dimensional sample complexity of MMD, which comes with unbiased gradient estimates. Learning Generative Models with the Sinkhorn Loss
Paul PEGON, Filippo SANTAMBROGIO, Qinglan XIA
On s'intéresse à la question suivante : quel est l'ensemble de volume un qui s'irrigue le mieux à partir d'une source ponctuelle située à l'origine, au sens du transport branché ? Nous pouvons formuler cette question comme un problème d'optimisation de forme et prouver l'existence de solutions, qui peuvent être considérées comme des sortes de "boule unité" pour le transport branché. La régularité $\beta$-Hölder de la fonction paysage permet d'obtenir une borne supérieure sur la dimension de Minkowski du bord : $\overline{\dim}_M \partial A \leq d-\beta$ (pour un certain exposant $\beta \in ]0,1[$), que l'on conjecture comme étant la dimension (non-entière) du bord. Nous proposons une première approche pour calculer numériquement une forme optimale approchée, utilisant une adaptation de l'approximation par champs de phase du transport branché introduite il y a quelques années par Oudet et Santambrogio. Afin de rechercher un minimiseur de la fonctionnelle approchée, on met en oeuvre des méthodes d'optimisation convexe non lisse, de type gradient et quasi-Newton proximal. Une forme optimale en transport branché
Pierre MARECHAL
L'approche variationnelle de la mollification s'écrit assez naturellement pour les opérateurs ayant une parenté avec l'opérateur de Fourier (convolution, opérateur de Radon). Dans le cas général, il est possible d'étendre le cadre via la notion de relation d'entrelacement d'opérateurs. Ceci permet non seulement d'étendre la classe des problèmes inverses pouvant être régularisés de cette manière, mais aussi d'envisager des approximations de l'unité plus générales que celles issues d'opérateurs de convolution. Cette communication a pour but de présenter le cadre général qui en résulte et de donner quelques résultats de convergence associés. Sur l'approche variationnelle de la régularisation par mollication
|
Structural properties of games (CA16228 GAMENET session)
Structural properties of games (CA16228 GAMENET session) Joseph ABDOU, Nikolaos PNEVMATIKOS, Marco SCARSINI, Xavier VENEL
In this work, we generalize the orthogonal decomposition of games presented in Candogan et al. (2011) to several metrics induced in the space of games. We in- troduce a family of inner products in the space of finite games and using the Helmholtz-Hodge decomposition theorem, we establish an orthogonal decomposition of finite games into three components, which we refer to as the potential, μ-harmonic and non-strategic components. We prove that a completely mixed strategy profile, related to the selected inner product, ap- pears as an equilibrium in the μ-harmonic component and that μ-harmonic games do not generically have pure Nash equilibria. We further show that in two-player μ-harmonic games, coincidence between the sets of correlated and mixed equilibria holds true and if players have equal number of actions, it then follows uniqueness of the equilibrium. Metrics and harmonicity in the space of games
Philippe BICH
A weighted network is defined by some agents, and for each non ordered pair of nodes some number measuring the strength of the relationship between them. Each agent has some preference on the set of networks, represented by a payoff function.
A network is pairwise stable if no agent has some interest to unilateraly decrease the weight of some of his link, and no two agents have some interest to
bilateraly increase the intensity of their relationship.
The concept has been intensively used in Network theory since 20 years to represent a stable network in any formation network process. But there is no general existence result in the spirit of Nash existence result. We give such result and also raise the following questions (and solve some of them):
- Can pairwise stable network be obtained as a Nash equilibrium of some "natural" game ?
- Is there some notion of mixed-strategy pairwise stable network, similar to game theory ?
- What can be said, in general, about the number of pairwise stable networks ?
- What are the natural dynamics related to pairwise stable networks ?
On the Existence and the number of Pairwise stable networks
Jean-Marie MIREBEAU, Johann DREO
Nous considérons un jeu à deux joueurs, où le premier joueur doit installer un système de surveillance,
sous la forme d'un ensemble de capteurs disposés dans une région admissible.
Le second joueur doit entrer dans la zone surveillée, visiter une région cible, puis quitter la zone, tout en minimisant sa probabilité globale de détection.
Les deux joueurs connaissent la région cible et le deuxième joueur connaît les détails de l'installation de surveillance.
Les trajectoires optimales pour le second joueur sont calculées en utilisant une variante récemment développée de l'algorithme de marche rapide,
prenant en compte une anisotropie de détection et des contraintes de courbure modélisant la maniabilité du véhicule du second joueur.
L'optimisation du système de surveillance par une méthode de quasi-Newton exploite une procédure de différenciation semi-automatique en mode inverse,
estimant le gradient de la fonction de valeur liée à l'emplacement du réseau de capteurs en temps O(N ln N).
Différentiation automatique d'une marche rapide non-holonomique pour le calcul de trajectoires menaçantes sous surveillance de capteurs
| ||||
12h-14h | Repas | |||||
14h-16h30 | Activités de groupe (biathlon-ludique, ski,... travail!) | |||||
16h30-18h00 |
Nonconvex optimization and applications
Nonconvex optimization and applications Victor MAGRON, Mohab SAFEY EL DIN
We consider the problem of finding exact sums of squares (SOS) decompositions for certain classes of nonnegative multivariate polynomials, while relying on semidefinite programming (SDP) solvers.
We start by providing a hybrid symbolic-numeric algorithm computing exact rational SOS decompositions for polynomials lying in the interior of the SOS cone. This algorithm computes an approximate SOS decomposition for a perturbation of the input polynomial with an arbitrary-precision SDP solver. An exact SOS decomposition is obtained thanks to the perturbation terms.
We prove that bit complexity estimates on output size and runtime are both polynomial in the degree of the input polynomial and simply exponential in the number of variables.
This analysis is based on quantifier elimination as well as bounds on the cost of the ellipsoid method and Cholesky's decomposition.
Next, we apply this algorithm to compute exact Polya and Putinar's representations respectively for positive definite forms and polynomials positive over basic compact semialgebraic sets.
We also compare the implementation of our algorithms with existing methods based on CAD or critical points. On Exact Polynomial Optimization
Samir ADLY
Dans cet exposé, nous étudions une nouvelle variante du processus de rafle de Moreau avec contrainte sur la vitesse. En utilisant une version adaptée de l'algorithme de rattrapage de Moreau, nous montrons l'existence et l'unicité de solution. Nous montrons aussi l'équivalence entre ce processus implicite et une évolution quasistatique d'une inéquation variationnelle.
Il est bien connu que la formulation variationnelle de nombreux problèmes mécaniques avec contact unilatéral et frottement conduit à une inéquation variationnelle d'évolution. Comme application, nous reformulons le problème de contact quasistatique avec frottement pour les matériaux élastiques linéaires à mémoire courte en tant que processus de Moreau implicite avec contrainte sur la vitesse. Le lien entre le processus de Moreau implicite et les inéquations variationnelles quasistatiques est possible grâce à certains outils standards de l'analyse convexe. Problèmes de rafle de Moreau en mécanique unilatérale
|
Controlled Dynamical Systems
Controlled Dynamical Systems Vincenzo BASCO
In this talk I will discuss sufficient conditions for Lipschitz regularity of the value function for an infinite horizon optimal control problem subject to state constraints. I focus on problems with cost functional admitting a discount rate factor and allow time dependent dynamics and lagrangian. Furthermore, state constraints may be unbounded and may have a nonsmooth boundary. Lipschitz regularity is recovered as a consequence of estimates on the distance of a given trajectory of control system from the set of all its viable (feasible) trajectories, provided the discount rate is sufficiently large. As the first application it is shown that the value function of the original problem coincides with the value function of the relaxed infinite horizon problem. The second application concerns first order necessary optimality conditions: a constrained maximum principle and sensitivity relations involving generalized gradients of the value function. Necessary Conditions and Lipschitz Continuity of the Value Function for Infinite Horizon optimal control problems under State Constraints
Chadi NOUR
Under suitable assumptions, the epigraph of the bilateral minimal function is proved to be $\varphi$-convex for a nonlinear control system. This generalized, to the nonlinear case, the main result of [2] where a similar result is proved for a linear control system. Semiconvexity of the bilateral minimal time function: The nonlinear case.
Lamberto DELL'ELCE, Jean-Baptiste CAILLAU, Jean-Baptiste POMET
Averaging is a valuable technique to gain understanding in the long-term evolution of dynamical systems characterized by slow and fast dynamics. Short period variations of averaged trajectories can be restored a posteriori by means of a near-identity transformation that is a function of both the averaged slow and fast variables. Recent contributions in optimal control theory prove that averaging can be applied to the dynamical system resulting from the necessary conditions for optimality. The present talk extends these results by discussing the evaluation of short-period variations of the adjoint variables. First, the classical approach is shown to be inadequate when applied to the assessment of the adjoints of slow variables because of the peculiar form of their equations of motion. Hence, a consistent transformation is developed, such that variations of the adjoints of fast and slow variables are evaluated in sequence. A simplified transformation is finally obtained when a single fast variable is considered. The methodology is applied to a time-optimal low-thrust orbital transfer. Restoring Short-Period Oscillations of the Motion of Averaged Optimal Control Systems
| ||||
18h00-18h15 | Pause | |||||
18h15-19h15 | Alberto BRESSAN Conservation law models for traffic flow on networks Traffic flow on a network of roads can be modeled by a family of conservation laws, describing the density of cars along each road. These are supplemented by suitable boundary conditions, modeling flow at intersections. In addition, one can introduce a cost functional, accounting for the time that each driver spends on the road and a penalty for late arrival. This leads to the problem of finding a globally optimal solution, which minimizes the sum of the costs to all drivers, or a Nash equilibrium solution, where no driver can lower his individual cost by changing his own departure time or his route to reach destination. This talk will survey some recent results on the well-posedness of the PDE models, necessary conditions for global optimality, and the characterization of Nash equilibria. Some open questions will also be discussed. Conservation law models for traffic flow on networks Modérateur: Jean-Baptiste CAILLAU, salle plénière (Vercors) | |||||
19h30-21h30 | Dîner festif | |||||
21h30-22h30 | AG du groupe |
9h-10h | Sylvain SORIN Optimization, learning and games: replicator dynamics, old and new We introduce the unilateral version associated to the replicator dynamics and its connection to on-line learning procedures and classical gradient algorithms in discrete and continuous time. We will survey recent results on extensions of this dynamics: regularization functions and variable weights, time average and link to best reply dynamics in games, equilibria and variational inequalities, potential and dissipative games, Hessian Riemannian metrics. Optimization, learning and games: replicator dynamics, old and new Modérateur: Rida LARAKI, salle plénière (Vercors) | |||||
10h-10h30 | Pause Café | |||||
10h30-12h30 |
Machine learning
Machine learning Alberto BIETTI, Julien MAIRAL
We study deep signal representations that are invariant to groups of transformations and stable to the action of diffeomorphisms without losing signal information. This is achieved by generalizing the multilayer kernel construction introduced in the context of convolutional kernel networks and by studying the geometry of the corresponding reproducing kernel Hilbert space. We show that the signal representation is stable, and that models from this functional space, such as a large class of convolutional neural networks with homogeneous activation functions, may enjoy the same stability. In particular, we study the norm of such models, which acts as a measure of complexity, controlling both stability and generalization. Group Invariance, Stability to Deformations, and Complexity of Deep Convolutional Representations
Aude GENEVAY
Generative models are a class of statistical models that can easily be sampled from but whose likelihood is intractable because they are supported on low-dimensional manifolds in much higher-dimensional spaces. Inference for these models thus can’t be performed with standard methods using maximum likelihood. That’s why the ability to compare two degenerate probability distributions is a crucial factor in the estimation of these generative models. It is therefore no surprise that optimal transport (OT) metrics and their ability to handle measures with non-overlapping supports have emerged as a promising tool.
Yet, training generative models using OT raises formidable computational and statistical challenges, because of (i) the computational burden of eval- uating OT losses, (ii) their instability and lack of smoothness, (iii) the difficulty to estimate them, as well as their gradients, in high dimension. By introducing an OT-based loss called Sinkhorn loss, we can tackle these three issues by relying on two key ideas: (a) entropic smoothing, which turns the original OT loss into a differentiable and more robust quantity that can be computed using Sinkhorn fixed point iterations; (b) algorithmic (automatic) differentiation of these iterations with seamless GPU execution. Additionally, Entropic smoothing generates a family of losses interpolating between Wasserstein (OT) and Energy distance/Maximum Mean Discrepancy (MMD) losses, thus allowing to find a sweet spot leveraging the geometry of OT on the one hand, and the favorable high-dimensional sample complexity of MMD, which comes with unbiased gradient estimates. Learning Generative Models with the Sinkhorn Loss
Cyprien GILET, Marie DEPREZ, Jean-Baptiste CAILLAU, Michel BARLAUD
Ce travail traite de classification non supervisée avec sélection de variables. Le
problème est d'estimer à la fois les labels ainsi qu'une matrice creuse de poids pour
réduire la dimension des données. Pour traiter ce problème combinatoire non
convexe tout en maintenant le contrôle de la parcimonie de la matrice de poids, nous
proposons une minimisation alternée de la norme de Frobenius de notre critère.
L'algorithme qui en résulte, "k-sparse", alterne k-means et gradient projeté.
L'étape de gradient projeté est de type splitting, avec une projection exacte sur la
boule $\ell^1$ pour assurer la parcimonie. La convergence de l'étape de gradient
projeté est établie. La norme de Frobenius de notre critère converge lorsque le nombre
d'itérations de notre algorithme tend vers l'infini. Plusieurs tests sur des bases de
données Single Cell RNA-seq montrent que notre méthode améliore de façon significative
les résultats d'autres méthodes de clustering. La complexité de notre
algorithme est linéaire avec le nombre d'observations (cellules), de sorte que la
méthode est applicable sur de grandes bases de données. Clustering avec sélection de variables par minimisation alternée et application à la biologie
Henri GERARD, Michel DE LARA, Jean-Philippe CHANCELIER
Consider a finite number of agents, each of them endowed with their own preference on their own decision set.
A leader supervises the team of agents. He is endowed with his own preference over the product of agents decision sets.
We extend the notion of time consistency to one of agent consistency:
when the agents, one by one, are indifferent between two decisions, so is the team leader
with respect to the two collections of decisions.
When preferences are represented by objective functions, we prove that agent consistency
is equivalent to a nested formula: the team leader objective function is a function (aggregator) of the agents objective functions.
In other words, the team leader objective function can be factorized into the agents objective functions.
We also introduce a notion of strong agent consistency that goes beyond indifference,
and prove that the aggregator function is monotone.
We discuss applications to risk measures. From time consistency to agent consistency in stochastic optimization
|
Variational Analysis
Variational Analysis Pascal GOURDEL, Nadia MÂAGLI
In a previous paper, we showed that if $X$ is a metric space and $Y$ is a Banach space,
then any lower semicontinuous correspondence
$\varphi: X\to 2^Y$ with nonempty convex valued such that $\varphi$ has either closed or finite dimensional images
admits a selection. In a new paper, we extend this result to the case where $X$ is
Hausdorff paracompact and perfectly normal topological space.
This allows to revisit the pioneer work of Michael (956) and to show that such a property
is a characterization of Hausdorff paracompact and perfectly normal topological space.
As in [Gourdel-Maagli 2017], we use the concept of peeling for the points $x$ such that $\varphi(x)$ has a finite dimension
in order to build a lower semicontinuous correspondence
such that $\overline{\psi_{\eta}(x)}\subset {\mathop{\rm ri}}( \varphi(x))$.
In this paper, additional techniques are used to encompass the absence of a metric structure. A convex selection theorem with a non separable Banach space
Florent NACRY, Lionel THIBAULT
Cet exposé est consacré à la régularisation des processus d'évolution de Moreau, convexes et non-convexes, continus et discontinus, dans un Hilbert quelconque. Nous passerons en revue diverses contributions majeures de la littérature de 1971 à nos jours : J.J. Moreau dans le cas convexe absolument continu, M.D.P. Monteiro Marques lorsque la variation de l'ensemble est simplement à variation bornée puis la première étude non-convexe/prox-régulière due à L. Thibault ainsi que la récente régularisation de A. Jourani et E. Vilches non-prox-régulière/$\alpha$-far. Nous présenterons également nos nouveaux résultats pour un ensemble mobile prox-régulier dont le mouvement est contrôlé seulement à travers la distance de Pompeiu-Hausdorff tronquée. Régularisation des processus d'évolution de Moreau
Hassan SAOUD
We are interested in the systems of the form
$$\dot{x}(t) + F(x(t)) \in - \partial \varphi (x(t)),$$
Where $F : \varphi : \mathbb{R}^n \to \mathbb{R}^n$ is a continuous function and $\varphi : \mathbb{R}^n \to \mathbb{R} \cup \{+\infty\}$ is a proper, convex and lower
semi-continuous function.
The aim of this talk is to consider an alternative notion of stability called semistability which, in certain sense, lies between stability and asymptotic stability.
More precisely, an equilibrium is semistable if it is Lyapunov stable, and every trajectory starting in a neighborhood of the equilibrium converges to a (possibly different) Lyapunov stable equilibrium.
It can be seen that, for an equilibrium, asymptotic stability implies semistability, while semistability implies Lyapunov stability. In addition to semistability,
it is desirable that a dynamical system that exhibits semistability also possesses the property that trajectories that converge
to a Lyapunov stable system state must do so in finite time rather than merely asymptotically. This is the so called finite-time semistability.\\
Two approaches will be used to treat these two notions of convergence and semistability. In the first one, the results are Lyapunov-based and are obtained without any assumptions of sign definiteness on the Lyapunov function.
In the second one, we use a condition based on nontangency between the vector field and invariant or negatively invariant subsets of the level or sublevel sets of the Lyapunov function or its
derivative. Semistability Theory of Nonsmooth Dynamical Systems
| ||||
12h30-13h30 | Buffet (ou panier repas) | |||||
13h45-15h | Navettes de retour vers la gare de Grenoble |