Abstract:Networks capture pairwise interactions between entities and are frequently used in applications such as social networks, food networks, and protein interaction networks, to name a few. Communities, cohesive groups of nodes, often form in these applications, and identifying them gives insight into the overall organization of the network. One common quality function used to identify community structure is modularity. In Hu et al. [SIAM J. App. Math., 73(6), 2013], it was shown that modularity optimization is equivalent to minimizing a particular nonconvex total variation (TV) based functional over a discrete domain. They solve this problem, assuming the number of communities is known, using a Merriman, Bence, Osher (MBO) scheme. We show that modularity optimization is equivalent to minimizing a convex TV-based functional over a discrete domain, again, assuming the number of communities is known. Furthermore, we show that modularity has no convex relaxation satisfying certain natural conditions. We therefore, find a manageable non-convex approximation using a Ginzburg Landau functional, which provably converges to the correct energy in the limit of a certain parameter. We then derive an MBO algorithm with fewer hand-tuned parameters than in Hu et al. and which is 7 times faster at solving the associated diffusion equation due to the fact that the underlying discretization is unconditionally stable. Our numerical tests include a hyperspectral video whose associated graph has 2.9x10^7 edges, which is roughly 37 times larger than was handled in the paper of Hu et al.

Mapping The Energy Landscape Of Non-Convex Optimization Problems

Mapping Energy Landscapes of Non-Convex Learning Problems.

Deep Manifold Computing and Visualization Using Elastic Locally Isometric Smoothness

Perspective: Energy Landscapes for Machine Learning

Simplified Energy Landscape for Modularity Using Total Variation

Spatial-domain fitness landscape analysis for combinatorial optimization

Learning Clustered Representation for Complex Free Energy Landscapes

ESSP: an Efficient Approach to Minimizing Dense and Nonsubmodular Energy Functions

Distributed Learning with Convex SUM-of -Non-convex Objective

Minimizing Sparse Higher Order Energy Functions of Discrete Variables

Clustering Molecular Energy Landscapes by Adaptive Network Embedding

Deep Representation Learning for Complex Free Energy Landscapes.

Energy landscape analysis of neuroimaging data

Geometric Manifold Energy and Manifold Clustering

Energy landscape analysis based on the Ising model: Tutorial review

An Improved Energy Landscape Paving Algorithm for the Problem of Packing Circles into a Larger Containing Circle.

Energy-Based Learning for Cooperative Games, with Applications to Valuation Problems in Machine Learning

Visualizing Basins of Attraction for Different Minimization Algorithms

Energy landscapes and rare events

A Comparative Study of Modern Inference Techniques for Discrete Energy Minimization Problems

Energy minimization and relaxation labeling