Geometry of the restricted Boltzmann machine
Maria Angelica Cueto, Jason Morton, Bernd Sturmfels

TL;DR
This paper explores the algebraic and geometric properties of restricted Boltzmann machines, revealing their structure through tropical geometry and algebraic statistics, and establishing conditions for their identifiability.
Contribution
It introduces a novel algebraic geometric framework for analyzing RBMs, including dimension formulas and identifiability results, connecting them to secant varieties and tropical geometry.
Findings
Zariski closure as Hadamard power of secant variety
Dimension formula for tropicalized RBM
Identifiability in many cases
Abstract
The restricted Boltzmann machine is a graphical model for binary random variables. Based on a complete bipartite graph separating hidden and observed variables, it is the binary analog to the factor analysis model. We study this graphical model from the perspectives of algebraic statistics and tropical geometry, starting with the observation that its Zariski closure is a Hadamard power of the first secant variety of the Segre variety of projective lines. We derive a dimension formula for the tropicalized model, and we use it to show that the restricted Boltzmann machine is identifiable in many cases. Our methods include coding theory and geometry of linear threshold functions.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsModel Reduction and Neural Networks · Theoretical and Computational Physics · Lattice Boltzmann Simulation Studies
MethodsRestricted Boltzmann Machine
