The Rise of Federated Intelligence: From Federated Foundation Models Toward Collective Intelligence
The Rise of Federated Intelligence: From Federated Foundation Models Toward Collective Intelligence
Guodong Long
Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence
Early Career. Pages 8547-8552.
https://doi.org/10.24963/ijcai.2024/980
The success of foundation models advances the development of various intelligent and personalized agents to handle intricate tasks in their daily lives, however finite resources and privacy concerns from end users limit the potential of customizing the large intelligent agents for personal use. This paper explores the preliminary design of federated intelligence that paves the way toward personalized intelligent agents in large-scale collaboration scenarios. In Federated Intelligence, agents can collaboratively augment their intelligence quotient (IQ) by learning complementary knowledge and fine-grained adaptations. These personalized intelligent agents can also co-work together to jointly address complex tasks in the form of collective intelligence. The paper will highlight federated intelligence as a new pathway for tackling complex intelligent tasks by refining and extending centralized foundation models to an open and collaborative paradigm.
Keywords:
Machine Learning: ML: Federated learning
Agent-based and Multi-agent Systems: MAS: Coordination and cooperation