Manifold Learning for Visualizing and Analyzing High-dimensional Data

Assuming that high-dimensional data are generated from intrinsic variables with lower dimensions, several key manifold-learning algorithms can help effectively analyze and visualize such data.

Authors

• Junping Zhang
• Hua Huang Beijing Normal University
• Jue Wang

Keywords

• Data visualization
• Data analysis
• Principal component analysis
• Machine learning
• Pixel
• IEEE members
• Statistical analysis
• Statistics
• Manifolds
• Machine learning algorithms
• Dimensionality Reduction
• Image Pixels
• Euclidean Space
• Intrinsic Variability
• Linear Projection
• Linear Method
• Low-dimensional Space
• Orthogonal Matrix
• Translation Invariance
• Neighboring Points
• Graph Laplacian
• Spectral Decomposition
• Diffusion Maps
• Geodesic Distance
• Locally Linear Embedding
• Semidefinite Programming
• Linear Subspace
• Gram Matrix
• Low-dimensional Subspace
• Norm Constraint
• Kernel Principal Component Analysis
• Low-dimensional Manifold
• Heat Kernel
• Swiss Roll
• Dimension Estimation
• Intrinsic Structure
• Column Vector
• Shortest Path
• Long-term Memory
• Facial Expressions
• manifold learning
• multivariate statistics
• pattern analysis
• intelligent systems