Information-Geometric Set Embeddings (IGSE): From Sets to Probability Distributions
This work addresses a foundational problem in machine learning for representing sets, but it is incremental as it only provides a preliminary solution on toy data.
The paper tackles the problem of embedding sets into probability distributions to minimize information loss, relating set operations to distribution interpolations and demonstrating a preliminary solution with experimental results on toy examples.
This letter introduces an abstract learning problem called the "set embedding": The objective is to map sets into probability distributions so as to lose less information. We relate set union and intersection operations with corresponding interpolations of probability distributions. We also demonstrate a preliminary solution with experimental results on toy set embedding examples.