Automatically Composing Representation Transformations as a Means for Generalization

Michael B. Chang, Abhishek Gupta, Sergey Levine, Tom Griffiths

Solving a problem simply means representing it so as to make the solution transparent.

-- Herbert Simon, 1988

Question: how to build machines that leverage prior experience to solve different and more complex problems than they have previously encountered?

Compositional Recursive Learner: a domain-general framework for learning algorithmic procedures for composing representation transformations, producing a learner that reasons about what computation to execute by making analogies to previously seen problems.

@article{chang2018automatically,
  title={Automatically composing representation transformations as a means for generalization},
  author={Chang, Michael B and Gupta, Abhishek and Levine, Sergey and Griffiths, Thomas L},
  journal={arXiv preprint arXiv:1807.04640},
  year={2018}
}