Social intelligence in multi-agent deep reinforcement learning
Lead Research Organisation:
University College London
Department Name: Electronic and Electrical Engineering
Abstract
Social learning is a key component of intelligence. Evidence from psychology suggests that in- dividuals of many animal species exploit the experience of others by learning from them [1]. Nonetheless, the role of social behaviour in artificial intelligence is currently marginalised. This research proposal brings together cognitive psychology, neuroscience, and computer science to focus on how we can endow artificial agents with elements of social intelligence.
This research project is aimed at answering the following questions: "Can the ability of artificial intelligence to learn, adapt, and generalise to new environments be enhanced by mechanisms that allow for social learning?". Also, "Can the interaction of multiple agents generate forms of specialisation? Do they learn hierarchies? Do they learn the value of diversity?". And finally, "Can we model the dynamics of human relationships?", "Can we model our inner self and the relationship with our behaviour toward others?".
EPSRC thematic areas ICT - Artificial Intelligence
This research project is aimed at answering the following questions: "Can the ability of artificial intelligence to learn, adapt, and generalise to new environments be enhanced by mechanisms that allow for social learning?". Also, "Can the interaction of multiple agents generate forms of specialisation? Do they learn hierarchies? Do they learn the value of diversity?". And finally, "Can we model the dynamics of human relationships?", "Can we model our inner self and the relationship with our behaviour toward others?".
EPSRC thematic areas ICT - Artificial Intelligence
Organisations
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
EP/R513143/1 | 01/10/2018 | 30/09/2023 | |||
2434391 | Studentship | EP/R513143/1 | 02/11/2020 | 01/11/2024 | Eduardo Pignatelli |
EP/T517793/1 | 01/10/2020 | 30/09/2025 | |||
2434391 | Studentship | EP/T517793/1 | 02/11/2020 | 01/11/2024 | Eduardo Pignatelli |