Eigenvalue bounds for an alignment matrix in manifold learning (Q413545)

From the authors' abstract: This paper presents a spectral analysis for an alignment matrix that arises in the reconstruction of a global coordinate system from local coordinate systems through alignment in manifold learning. Some characterizations of its eigenvalues and its null space as well as a lower bound for the smallest positive eigenvalue are given, which generalize earlier results of \textit{C.-K. Li}, \textit{R.-C. Li} and the first-named author [Commun. Math. Sci.~5, No. 2, 313--329 (2007; Zbl 1135.65019)] to include a more general situation that arises in alignments of local sections of different dimensions. The results provide a theoretical understanding of the local tangent space alignment method due to \textit{Z. Zhang} and \textit{H. Zha} [SIAM J. Sci. Comput. 26, No. 1, 313--338 (2004; Zbl 1077.65042)] for nonlinear manifold learning and address some computational issues related to the method.