Machine Learning on Sets
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact jg801.
For modality such as image and sequence(text, audio), the input/output ordering contains information which will lost after performing random permutation on the data. However, in several other domains such as (sub) graphs and 3D meshes/point clouds, it is more natural to represent each instance as the set of its components or parts. Many conventional machine learning algorithms are unable to process this kind of representations, since sets may vary in cardinality and elements lack a meaningful ordering. In this talk, we will present a paper introducing a framework called ‘Deep Sets’ to deal with such set representations, whose permutation invariance has been proved necessarily and sufficiently. Furthermore, we will discuss some specific permutation invariant/equivariant (model & loss) designs on 3D Vision and Graphs.
This talk is part of the Machine Learning Reading Group @ CUED series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|