KAMA-NNs: Low-dimensional Rotation Based Neural Networks.
Krzysztof ChoromanskiAldo PacchianoJeffrey PenningtonYunhao TangPublished in: AISTATS (2019)
Keyphrases
- neural network
- low dimensional
- high dimensional
- back propagation
- manifold learning
- dimensionality reduction
- artificial neural networks
- principal component analysis
- input space
- training algorithm
- fuzzy logic
- high dimensional data
- multilayer perceptron
- recurrent neural networks
- pattern recognition
- data points
- multi layer perceptron
- feature space
- neural nets
- activation function
- learning capabilities
- genetic algorithm
- neural network model
- feed forward
- fault diagnosis
- neyman pearson
- knn
- nearest neighbor
- fuzzy systems
- euclidean space
- rotation invariant
- feedforward neural networks
- neural network training
- radial basis function
- self organizing maps
- invariant features
- hidden layer
- learning rules
- radial basis function neural network
- multidimensional scaling
- hybrid systems
- embedding space