Descripción del proyecto
Ampliación del aprendizaje por refuerzo teóricamente sólido
El aprendizaje por refuerzo (AR) es un subcampo del aprendizaje automático que se ocupa de cómo los agentes inteligentes interactúan con entornos desconocidos para maximizar sus recompensas. La aplicación potencial de las técnicas de AR en problemas difíciles reales, como el control de vehículos autónomos o las redes de energía inteligentes, ha atraído una atención considerable a este campo. Sin embargo, los algoritmos de AR más avanzados no se pueden aplicar en los dominios más prometedores, en gran parte debido a la falta de garantías formales de rendimiento. El proyecto SCALER, financiado con fondos europeos, pretende abordar este reto adoptando un enfoque basado en principios para desarrollar una nueva generación de algoritmos de AR con una eficiencia y una escalabilidad demostrables. La metodología se basará en la identificación de nuevas propiedades estructurales de los procesos de decisión de Markov a gran escala que permitan un aprendizaje eficiente desde el punto de vista computacional y estadístico.
Objetivo
Reinforcement learning (RL) is an intensely studied subfield of machine learning concerned with sequential decision-making problems where a learning agent interacts with an unknown reactive environment while attempting to maximize its rewards. In recent years, RL methods have gained significant popularity due to being the key technique behind some spectacular breakthroughs of artificial intelligence (AI) research, which renewed interest in applying such techniques to challenging real-world problems like control of autonomous vehicles or smart energy grids. While the RL framework is clearly suitable to address such problems, the applicability of the current generation of RL algorithms is limited by a lack of formal performance guarantees and a very low sample efficiency. This project proposes to address this problem and advance the state of the art in RL by developing a new generation of provably efficient and scalable algorithms. Our approach is based on identifying various structural assumptions for Markov decision processes (MDPs, the main modeling tool used in RL) that enable computationally and statistically efficient learning. Specifically, we will focus on MDP structures induced by various approximation schemes including value-function approximation and relaxations of the linear-program formulation of optimal control in MDPs. Based on this view, we aim to develop a variety of new tools for designing and analyzing RL algorithms, and achieve a deep understanding of fundamental performance limits in structured MDPs. While our main focus will be on rigorous theoretical analysis of algorithms, most of our objectives are inspired by practical concerns, particularly by the question of scalability. As a result, we expect that our proposed research will have significant impact on both the theory and practice of reinforcement learning, bringing RL methods significantly closer to practical applicability.
Palabras clave
Programa(s)
Régimen de financiación
ERC-STG - Starting GrantInstitución de acogida
08002 Barcelona
España