Speeding-up continual learning through information gains in novel experiences

摘要: Adapting to novelties in open-world environments is an important and difficult challenge, and it has been recently shown that hybrid planning and reinforcement learning approaches …