A Deep Neural Network Architecture Using Dimensionality Reduction with Sparse Matrices.
Wataru MatsumotoManabu HagiwaraPetros T. BoufounosKunihiko FukushimaToshisada MariyamaXiongxin ZhaoPublished in: ICONIP (4) (2016)
Keyphrases
- sparse matrices
- dimensionality reduction
- floating point
- singular value decomposition
- high dimensional
- pattern recognition
- linear algebra
- high dimensional data
- principal component analysis
- high dimensionality
- feature extraction
- data representation
- data points
- low dimensional
- condition number
- feature space
- principal components
- random projections
- rows and columns
- feature selection
- sparse representation
- original data
- metric learning
- kernel function