Publication

A new regret analysis for Adam-type algorithms

Publications associées (29)

Momentum-Based Policy Gradient with Second-Order Information

Patrick Thiran, Saber Salehkaleybar, Negar Kiyavash, Mohammadsadegh Khorasani

Variance-reduced gradient estimators for policy gradient methods have been one of the main focus of research in the reinforcement learning in recent years as they allow acceleration of the estimation process. We propose a variance-reduced policy-gradient m ...

2022

Optimization and equilibrium problems with discrete choice models

Stefano Bortolomiol

Many transportation markets are characterized by oligopolistic competition. In these markets customers, suppliers and regulators make decisions that are influenced by the preferences and the decisions of all other agents. In particular, capturing and under ...

EPFL2022

Optimization Over Banach Spaces: A Unified View on Supervised Learning and Inverse Problems

Shayan Aziznejad

In this thesis, we reveal that supervised learning and inverse problems share similar mathematical foundations. Consequently, we are able to present a unified variational view of these tasks that we formulate as optimization problems posed over infinite-di ...

EPFL2022

Complexity analysis of stochastic gradient methods for PDE-constrained optimal control problems with uncertain parameters

Fabio Nobile, Matthieu Claude Martin, Sebastian Krumscheid

We consider the numerical approximation of an optimal control problem for an elliptic Partial Differential Equation (PDE) with random coefficients. Specifically, the control function is a deterministic, distributed forcing term that minimizes the expected ...

2021

Adaptation in Stochastic Algorithms: From Nonsmooth Optimization to Min-Max Problems and Beyond

Ahmet Alacaoglu

Stochastic gradient descent (SGD) and randomized coordinate descent (RCD) are two of the workhorses for training modern automated decision systems. Intriguingly, convergence properties of these methods are not well-established as we move away from the spec ...

EPFL2021

A dependence of the cost of fast controls for the heat equation on the support of initial datum

Hoài-Minh Nguyên

The controllability cost for the heat equation as the control time

T

goes to 0 is well-known of the order

e^{C/T}

for some positive constant

C

, depending on the controlled domain and for all initial datum. In this paper, we prove that the constant $C ...

2021

Optimal Thrust Vector Control of an Electric Small-Scale Rocket Prototype

Colin Neil Jones, Petr Listov, Roland Schwan

Recent advances in Model Predictive Control (MPC) algorithms and methodologies, combined with the surge of computational power of available embedded platforms, allows the use of real-time optimization-based control of fast mechatronic systems. This paper p ...

IEEE2022

Optimal Control Combining Emulation and Imitation to Acquire Physical Assistance Skills

Sylvain Calinon, Teguh Santoso Lembono, Amirreza Razmjoo Fard

paper studies exploiting action-level learning (imitation) in the optimal control problem context. Cost functions defined by the optimal control methods are similar to the goal-level learning (emulation) in animals. However, imitating the robot's or others ...

IEEE2021

STORM+: Fully Adaptive SGD with Momentum for Nonconvex Optimization

Volkan Cevher, Ali Kavis

In this work we investigate stochastic non-convex optimization problems wherethe objective is an expectation over smooth loss functions, and the goal is to find an approximate stationary point. The most popular approach to handling such problems is varianc ...

2021

Publications associées (29)

Complexity analysis of stochastic gradient methods for PDE-constrained optimal control problems with uncertain parameters

Fabio Nobile, Matthieu Claude Martin, Sebastian Krumscheid

2021

Adaptation in Stochastic Algorithms: From Nonsmooth Optimization to Min-Max Problems and Beyond

Ahmet Alacaoglu

EPFL2021

A dependence of the cost of fast controls for the heat equation on the support of initial datum

Hoài-Minh Nguyên

The controllability cost for the heat equation as the control time

T

goes to 0 is well-known of the order

e^{C/T}

for some positive constant

C

, depending on the controlled domain and for all initial datum. In this paper, we prove that the constant $C ...

2021