Improved Graph Laplacian via Geometric Self-Consistency
نویسندگان
چکیده
We address the problem of setting the kernel bandwidth used by Manifold Learning algorithms to construct the graph Laplacian. Exploiting the connection between manifold geometry, represented by the Riemannian metric, and the Laplace-Beltrami operator, we set by optimizing the Laplacian’s ability to preserve the geometry of the data. Experiments show that this principled approach is effective and robust.
منابع مشابه
Learning on Graph with Laplacian Regularization
We consider a general form of transductive learning on graphs with Laplacian regularization, and derive margin-based generalization bounds using appropriate geometric properties of the graph. We use this analysis to obtain a better understanding of the role of normalization of the graph Laplacian matrix as well as the effect of dimension reduction. The results suggest a limitation of the standa...
متن کاملLaplacian Energy of a Fuzzy Graph
A concept related to the spectrum of a graph is that of energy. The energy E(G) of a graph G is equal to the sum of the absolute values of the eigenvalues of the adjacency matrix of G . The Laplacian energy of a graph G is equal to the sum of distances of the Laplacian eigenvalues of G and the average degree d(G) of G. In this paper we introduce the concept of Laplacian energy of fuzzy graphs. ...
متن کاملFault Detection and Isolation of Multi-Agent Systems via Complex Laplacian
This paper studies the problem of fault detection and isolation (FDI) for multi-agent systems (MAS) via complex Laplacian subject to actuator faults. A planar formation of point agents in the plane using simple and linear interaction rules related to complex Laplacian is achieved. The communication network is a directed, and yet connected graph with a fixed topology. The loss of symmetry in the...
متن کاملOn net-Laplacian Energy of Signed Graphs
A signed graph is a graph where the edges are assigned either positive ornegative signs. Net degree of a signed graph is the dierence between the number ofpositive and negative edges incident with a vertex. It is said to be net-regular if all itsvertices have the same net-degree. Laplacian energy of a signed graph is defined asε(L(Σ)) =|γ_1-(2m)/n|+...+|γ_n-(2m)/n| where γ_1,...,γ_n are the ei...
متن کاملConvergence of Laplacian Eigenmaps
Geometrically based methods for various tasks of data analysis have attracted considerable attention over the last few years. In many of these algorithms, a central role is played by the eigenvectors of the graph Laplacian of a data-derived graph. In this paper, we show that if points are sampled uniformly at random from an unknown submanifold M of RN , then the eigenvectors of a suitably const...
متن کامل