DIRESA, a distance-preserving nonlinear dimension reduction technique based on regularized autoencoders.
Geert De PaepeLesley De CruzPublished in: CoRR (2024)
Keyphrases
- distance preserving
- nonlinear dimension reduction
- manifold learning
- low dimensional
- random projections
- original data
- high dimensional data
- denoising
- graph properties
- high dimensional
- least squares
- dimensionality reduction
- data sets
- locally linear embedding
- directed graph
- face recognition
- principal component analysis
- high dimensionality
- high resolution
- image classification
- machine learning