Learning Representations that Support Extrapolation
Taylor Webb,u00a0Zachary Dulberg,u00a0Steven Frankland,u00a0Alexander Petrov,u00a0Randall Ou2019Reilly,u00a0Jonathan Cohen
Extrapolation u2013 the ability to make inferences that go beyond the scope of oneu2019s experiences u2013 is a hallmark of human intelligence. By contrast, the generalization exhibited by contemporary neural network algorithms is largely limited to interpolation between data points in their training corpora. In this paper, we consider the challenge of learning representations that support extrapolation. We introduce a novel visual analogy benchmark that allows the graded evaluation of extrapolation as a function of distance from the convex domain defined by the training data. We also introduce a simple technique, temporal context normalization, that encourages representations that emphasize the relations between objects. We find that this technique enables a significant improvement in the ability to extrapolate, considerably outperforming a number of competitive techniques.
