Scholarlink

Artificial Intelligence (AI) is becoming more prevalent in industrial applications and, specifically, in advanced robotics, both industrial and mobile. This research work explores federated learning techniques to generate more robust and efficient learning models, combining the strengths of different existing reinforcement learning algorithms compensating for their less efficient aspects. To this end, reinforcement learning algorithms that represent the current state of the art have been selected, trained using standard evaluation environments, and then hybridized using federated learning techniques. Tests carried out show a better performance in terms of reward and specially in episode length of the federated algorithm with respect the original reinforcement learning approaches.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using Federated Learning Techniques to Generalize Reinforcement Learning Approaches

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Alberto Tellaeche Iglesias,
  • Ignacio Fidalgo Astorquia,
  • Juan-Ignacio Vázquez,
  • José Gaviria de la Puerta

摘要

Artificial Intelligence (AI) is becoming more prevalent in industrial applications and, specifically, in advanced robotics, both industrial and mobile. This research work explores federated learning techniques to generate more robust and efficient learning models, combining the strengths of different existing reinforcement learning algorithms compensating for their less efficient aspects. To this end, reinforcement learning algorithms that represent the current state of the art have been selected, trained using standard evaluation environments, and then hybridized using federated learning techniques. Tests carried out show a better performance in terms of reward and specially in episode length of the federated algorithm with respect the original reinforcement learning approaches.