Representing Data by a Mixture of Activated Simplices

TL;DR

This paper introduces a novel data representation model using a mixture of activated simplices, which are geometric structures on the unit sphere, enabling efficient modeling, interpretation, and state-of-the-art results on benchmark datasets.

Contribution

The paper proposes a new simplicial model for data representation that learns a convex hull boundary on the sphere, accommodating inhomogeneous data and providing interpretability and strong performance.

Findings

Achieves state-of-the-art results on 3D pose datasets.

Provides effective reconstruction and classification.

Supports realistic generative modeling.

Abstract

We present a new model which represents data as a mixture of simplices. Simplices are geometric structures that generalize triangles. We give a simple geometric understanding that allows us to learn a simplicial structure efficiently. Our method requires that the data are unit normalized (and thus lie on the unit sphere). We show that under this restriction, building a model with simplices amounts to constructing a convex hull inside the sphere whose boundary facets is close to the data. We call the boundary facets of the convex hull that are close to the data Activated Simplices. While the total number of bases used to build the simplices is a parameter of the model, the dimensions of the individual activated simplices are learned from the data. Simplices can have different dimensions, which facilitates modeling of inhomogeneous data sources. The simplicial structure is bounded --- this is appropriate for modeling data with constraints, such as human elbows can not bend more than 180 degrees. The simplices are easy to interpret and extremes within the data can be discovered among the vertices. The method provides good reconstruction and regularization. It supports good nearest neighbor classification and it allows realistic generative models to be constructed. It achieves state-of-the-art results on benchmark datasets, including 3D poses and digits.