Improving Performance in Reinforcement Learning by Breaking Generalization in Neural Networks

@inproceedings{Ghiassian2020ImprovingPI,
  title={Improving Performance in Reinforcement Learning by Breaking Generalization in Neural Networks},
  author={Sina Ghiassian and Banafsheh Rafiee and Yat Long Lo and Adam White},
  booktitle={AAMAS},
  year={2020}
}
  • Sina Ghiassian, Banafsheh Rafiee, +1 author Adam White
  • Published in AAMAS 2020
  • Mathematics, Computer Science
  • Reinforcement learning systems require good representations to work well. For decades practical success in reinforcement learning was limited to small domains. Deep reinforcement learning systems, on the other hand, are scalable, not dependent on domain specific prior knowledge and have been successfully used to play Atari, in 3D navigation from pixels, and to control high degree of freedom robots. Unfortunately, the performance of deep reinforcement learning systems is sensitive to hyper… CONTINUE READING

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 41 REFERENCES

    Human-level control through deep reinforcement learning

    VIEW 5 EXCERPTS
    HIGHLY INFLUENTIAL

    Deep Reinforcement Learning that Matters

    VIEW 1 EXCERPT

    Overcoming catastrophic forgetting in neural networks

    VIEW 1 EXCERPT

    Meta-Learning Representations for Continual Learning

    VIEW 2 EXCERPTS