Eitan Levin

Any-Dimensional Learning by Sampling

Eitan Levin, Venkat Chandrasekaran

Many machine learning models are defined for inputs of different sizes, such as point clouds containing different numbers of points, sequences of tokens of different lengths, and graphs on different numbers of nodes. Such models are trained on finitely-many examples of necessaril…Read more