A Principal Component Analysis Approach for Embedding Local Symmetries into Deep Learning Algorithms.
Pierre-Yves LagravePublished in: SAFECOMP Workshops (2020)
Keyphrases
- principal component analysis
- learning algorithm
- deep architectures
- subspace learning
- dimensionality reduction
- locality preserving projections
- principal components
- covariance matrix
- dimension reduction
- face recognition
- multidimensional scaling
- independent component analysis
- feature extraction
- discriminant analysis
- training procedure
- machine learning algorithms
- nonlinear dimensionality reduction
- graph embedding
- multi dimensional scaling
- linear discriminant analysis
- learning problems
- machine learning
- vector space
- face images
- singular value decomposition
- low dimensional
- active learning
- back propagation
- feature space
- deep learning
- training data
- supervised learning
- learning rate
- neural classifier
- reinforcement learning
- neural network
- pattern recognition
- image processing
- symmetry breaking
- dimension reduction methods
- high dimensional
- learning process
- data hiding
- learning scheme
- unsupervised learning
- training examples
- learning tasks