Issue No. 03 - March (2006 vol. 28)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/TPAMI.2006.56
Martin H.C. Law , IEEE
Anil K. Jain , IEEE
Understanding the structure of multidimensional patterns, especially in unsupervised cases, is of fundamental importance in data mining, pattern recognition, and machine learning. Several algorithms have been proposed to analyze the structure of high-dimensional data based on the notion of manifold learning. These algorithms have been used to extract the intrinsic characteristics of different types of high-dimensional data by performing nonlinear dimensionality reduction. Most of these algorithms operate in a "batch” mode and cannot be efficiently applied when data are collected sequentially. In this paper, we describe an incremental version of ISOMAP, one of the key manifold learning algorithms. Our experiments on synthetic data as well as real world images demonstrate that our modified algorithm can maintain an accurate low-dimensional representation of the data in an efficient manner.
Incremental learning, dimensionality reduction, ISOMAP, manifold learning, unsupervised learning.
A. K. Jain and M. H. Law, "Incremental Nonlinear Dimensionality Reduction by Manifold Learning," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 28, no. , pp. 377-391, 2006.