Corpus ID: 6550107

Modular Continual Learning in a Unified Visual Environment

@article{Feigelis2017ModularCL,
  title={Modular Continual Learning in a Unified Visual Environment},
  author={Kevin T. Feigelis and Blue Sheffer and Daniel L. K. Yamins},
  journal={ArXiv},
  year={2017},
  volume={abs/1711.07425}
}
  • Kevin T. Feigelis, Blue Sheffer, Daniel L. K. Yamins
  • Published in ICLR 2017
  • Computer Science, Biology, Mathematics
  • A core aspect of human intelligence is the ability to learn new tasks quickly and switch between them flexibly. Here, we describe a modular continual reinforcement learning paradigm inspired by these abilities. We first introduce a visual interaction environment that allows many types of tasks to be unified in a single framework. We then describe a reward map prediction scheme that learns new tasks robustly in the very large state and action spaces required by such an environment. We… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    31
    Twitter Mentions

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 32 REFERENCES

    Adam: A Method for Stochastic Optimization

    VIEW 1 EXCERPT
    HIGHLY INFLUENTIAL

    Count-Based Exploration with Neural Density Models

    • A. PixelCNN Hyper-parameters
    • 2017
    VIEW 2 EXCERPTS

    Improved multitask learning through synaptic intelligence

    VIEW 1 EXCERPT

    Learning to Reason: End-to-End Module Networks for Visual Question Answering

    VIEW 2 EXCERPTS

    Overcoming catastrophic forgetting in neural networks

    VIEW 2 EXCERPTS