Fahlman-Type Activation Functions Applied to Nonlinear PCA Networks Provide a Generalised Independent Component Analysis.
Mark GirolamiColin FyfePublished in: ICANNGA (1997)
Keyphrases
- independent component analysis
- principal component analysis
- independent components
- principle component analysis
- factor analysis
- activation function
- source separation
- signal processing
- blind source separation
- neural network
- dimensionality reduction
- principal components
- high dimensional
- computer vision
- independent components analysis
- linear discriminant analysis
- kernel pca
- feed forward
- probabilistic model
- artificial neural networks
- training data
- feature extraction
- image processing
- machine learning