Timezone: »
Making an adaptive prediction based on input is an important ability for general artificial intelligence. In this work, we step forward in this direction and propose a semi-parametric method, Meta-Neighborhoods, where predictions are made adaptively to the neighborhood of the input. We show that Meta-Neighborhoods is a generalization of k-nearest-neighbors. Due to the simpler manifold structure around a local neighborhood, Meta-Neighborhoods represent the predictive distribution p(y | x) more accurately. To reduce memory and computation overheads, we propose induced neighborhoods that summarize the training data into a much smaller dictionary. A meta-learning based training mechanism is then exploited to jointly learn the induced neighborhoods and the model. Extensive studies demonstrate the superiority of our method.
Author Information
Siyuan Shan (University of North Carolina at Chapel Hill)
Yang Li (UNC-Chapel Hill)
Junier Oliva (UNC - Chapel Hill)
More from the Same Authors
-
2020 Poster: Exchangeable Neural ODE for Set Modeling »
Yang Li · Haidong Yi · Christopher Bender · Siyuan Shan · Junier Oliva -
2019 Workshop: Sets and Partitions »
Nicholas Monath · Manzil Zaheer · Andrew McCallum · Ari Kobren · Junier Oliva · Barnabas Poczos · Ruslan Salakhutdinov -
2019 Poster: Meta-Curvature »
Eunbyung Park · Junier Oliva