Meta-Learning Survey Traces Path to DeepMind's Adaptive Agent
A new survey paper on arXiv formalizes meta-learning and meta-reinforcement learning, chronicling the algorithms leading to DeepMind's Adaptive Agent. The authors provide a task-based formalization and consolidate concepts needed to understand generalist approaches. The paper is categorized under Computer Science > Artificial Intelligence.
Key facts
- The survey formalizes meta-learning and meta-reinforcement learning.
- It traces the path to DeepMind's Adaptive Agent.
- The paper is on arXiv under Computer Science > Artificial Intelligence.
- Meta-learning enables rapid adaptation to new tasks with minimal data.
- The survey uses a task-based formalization.
- It consolidates essential concepts for understanding generalist approaches.
- The Adaptive Agent is a key focus of the survey.
- The paper chronicles landmark algorithms in the field.
Entities
Institutions
- DeepMind
- arXiv