To predict a set from a vector, use gradient descent to find a set the encodes to that vector.

[arxiv] [code] [poster 1] [poster 2]

Current approaches for predicting sets from feature vectors ignore the unordered nature of sets and suffer from discontinuity issues as a result. We propose a general model for predicting sets that properly respects the structure of sets and avoids this problem. With a single feature vector as input, we show that our model is able to auto-encode point sets, predict the set of bounding boxes of objects in an image, and predict the set of attributes of these objects.

    author        = {Yan Zhang and Jonathon Hare and Adam Pr\"ugel-Bennett},
    title         = {{Deep Set Prediction Networks}},
    booktitle     = {Advances in Neural Information Processing Systems},
    year          = {2019},
    eprint        = {1906.06565},
    url           = {},