Dusparic
International Keynote Speaker
Reinforcement learning for sustainability: adapting in large-scale heterogeneous urban environments.
Abstract: Reinforcement learning (RL), and in particular its combination with deep neural networks, has seen major breakthroughs in the recent years, most notably outperforming humans in Go and Atari games. RL use is also extensively investigated in a range of practical applications, including those within smart cities. However, existing algorithms still fall short of being suitable for a wider use in such complex environments. Over the past decade, my research has focused on developing techniques that enable the use of RL for optimization of resource distribution in urban environments, for example, smart energy grid and intelligent urban traffic networks. These city-scale infrastructures share properties with many other large-scale systems, i.e., are characterized by distributed control, heterogeneity, presence of multiple and often conflicting goals, reliance on diverse sources of information, and above all the need for continuous adaptation. In this talk I will discuss the issues in applying RL in such environments and present the techniques we have developed to enable multi-agent multi objective optimization, adaptation in non-stationary environments, continuous knowledge transfer via parallel transfer learning and state space adaptation. I will conclude the talk by discussing further challenges in enabling real-world RL deployments, including further development of algorithms to ensure full and seamless adaptivity and highlighting the need for software testing techniques for RL-based applications ensuring their safe and correct performance.
Conferencista nacional
Arbeláez
Visión Artificial Biomédica
Conferencista nacional
Nuñez
Machine Learning en acción: Un panorama de aplicaciones.
Conferencista nacional
Riascos
Un Tour en Helicóptero por la Inteligencia Artificial y sus Aplicaciones al Diseño de Políticas Públicas
Panelista Microsoft
Memorias
Sé participe de los mejores momentos del Foro.