Zobrazeno 1 - 10
of 207
pro vyhledávání: '"Sunehag P"'
Autor:
Leibo, Joel Z., Vezhnevets, Alexander Sasha, Diaz, Manfred, Agapiou, John P., Cunningham, William A., Sunehag, Peter, Haas, Julia, Koster, Raphael, Duéñez-Guzmán, Edgar A., Isaac, William S., Piliouras, Georgios, Bileschi, Stanley M., Rahwan, Iyad, Osindero, Simon
What is appropriateness? Humans navigate a multi-scale mosaic of interlocking notions of what is appropriate for different situations. We act one way with our friends, another with our family, and yet another in the office. Likewise for AI, appropria
Externí odkaz:
http://arxiv.org/abs/2412.19010
Cooperation in multi-agent learning (MAL) is a topic at the intersection of numerous disciplines, including game theory, economics, social sciences, and evolutionary biology. Research in this area aims to understand both how agents can coordinate eff
Externí odkaz:
http://arxiv.org/abs/2312.05162
Autor:
Sunehag, Peter, Vezhnevets, Alexander Sasha, Duéñez-Guzmán, Edgar, Mordach, Igor, Leibo, Joel Z.
Many environments contain numerous available niches of variable value, each associated with a different local optimum in the space of behaviors (policy space). In such situations it is often difficult to design a learning process capable of evading d
Externí odkaz:
http://arxiv.org/abs/2302.01180
Autor:
Agapiou, John P., Vezhnevets, Alexander Sasha, Duéñez-Guzmán, Edgar A., Matyas, Jayd, Mao, Yiran, Sunehag, Peter, Köster, Raphael, Madhushani, Udari, Kopparapu, Kavya, Comanescu, Ramona, Strouse, DJ, Johanson, Michael B., Singh, Sukhdeep, Haas, Julia, Mordatch, Igor, Mobbs, Dean, Leibo, Joel Z.
Multi-agent artificial intelligence research promises a path to develop intelligent technologies that are more human-like and more human-compatible than those produced by "solipsistic" approaches, which do not consider interactions between agents. Me
Externí odkaz:
http://arxiv.org/abs/2211.13746
Autor:
Leibo, Joel Z., Duéñez-Guzmán, Edgar, Vezhnevets, Alexander Sasha, Agapiou, John P., Sunehag, Peter, Koster, Raphael, Matyas, Jayd, Beattie, Charles, Mordatch, Igor, Graepel, Thore
Publikováno v:
In International Conference on Machine Learning 2021 (pp. 6187-6199). PMLR
Existing evaluation suites for multi-agent reinforcement learning (MARL) do not assess generalization to novel situations as their primary objective (unlike supervised-learning benchmarks). Our contribution, Melting Pot, is a MARL evaluation suite th
Externí odkaz:
http://arxiv.org/abs/2107.06857
The challenge of developing powerful and general Reinforcement Learning (RL) agents has received increasing attention in recent years. Much of this effort has focused on the single-agent setting, in which an agent maximizes a predefined extrinsic rew
Externí odkaz:
http://arxiv.org/abs/2006.06051
Autor:
Leibo, Joel Z., Perolat, Julien, Hughes, Edward, Wheelwright, Steven, Marblestone, Adam H., Duéñez-Guzmán, Edgar, Sunehag, Peter, Dunning, Iain, Graepel, Thore
Here we explore a new algorithmic framework for multi-agent reinforcement learning, called Malthusian reinforcement learning, which extends self-play to include fitness-linked population size dynamics that drive ongoing innovation. In Malthusian RL,
Externí odkaz:
http://arxiv.org/abs/1812.07019
Autor:
Sunehag, Peter, Lever, Guy, Gruslys, Audrunas, Czarnecki, Wojciech Marian, Zambaldi, Vinicius, Jaderberg, Max, Lanctot, Marc, Sonnerat, Nicolas, Leibo, Joel Z., Tuyls, Karl, Graepel, Thore
We study the problem of cooperative multi-agent reinforcement learning with a single joint reward signal. This class of learning problems is difficult because of the often large combined action and observation spaces. In the fully centralized and dec
Externí odkaz:
http://arxiv.org/abs/1706.05296
Autor:
Dulac-Arnold, Gabriel, Evans, Richard, van Hasselt, Hado, Sunehag, Peter, Lillicrap, Timothy, Hunt, Jonathan, Mann, Timothy, Weber, Theophane, Degris, Thomas, Coppin, Ben
Being able to reason in an environment with a large number of discrete actions is essential to bringing reinforcement learning to a larger class of problems. Recommender systems, industrial plants and language models are only some of the many real-wo
Externí odkaz:
http://arxiv.org/abs/1512.07679
Autor:
Sunehag, Peter, Evans, Richard, Dulac-Arnold, Gabriel, Zwols, Yori, Visentin, Daniel, Coppin, Ben
Many real-world problems come with action spaces represented as feature vectors. Although high-dimensional control is a largely unsolved problem, there has recently been progress for modest dimensionalities. Here we report on a successful attempt at
Externí odkaz:
http://arxiv.org/abs/1512.01124