Learning Representations that Support Extrapolation
Taylor Webb, Zachary Dulberg, Steven Frankland, Alexander Petrov, Randall O’Reilly, Jonathan Cohen
Abstract
Extrapolation – the ability to make inferences that go beyond the scope of one’s experiences – 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.
BibTeX
@InProceedings{pmlr-v119-webb20a,
title = {Learning Representations that Support Extrapolation},
author = {Webb, Taylor and Dulberg, Zachary and Frankland, Steven and Petrov, Alexander and O'Reilly, Randall and Cohen, Jonathan},
booktitle = {Proceedings of the 37th International Conference on Machine Learning},
pages = {10136--10146},
year = {2020},
editor = {III, Hal Daumé and Singh, Aarti},
volume = {119},
series = {Proceedings of Machine Learning Research},
month = {13--18 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v119/webb20a/webb20a.pdf},
url = {https://proceedings.mlr.press/v119/webb20a.html},
abstract = {Extrapolation – the ability to make inferences that go beyond the scope of one’s experiences – 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.}
}