Noise Stability Is Computable and Approximately Low-Dimensional.
Anindya DeElchanan MosselJoe NeemanPublished in: Computational Complexity Conference (2017)
Keyphrases
- low dimensional
- high dimensional
- dimensionality reduction
- high dimensional data
- manifold learning
- noise level
- data points
- input space
- missing data
- dimension reduction
- noisy data
- random noise
- signal to noise ratio
- noise model
- gaussian noise
- principal component analysis
- stability analysis
- nonlinear dimensionality reduction
- data sets
- linear subspace
- linear dimensionality reduction
- arbitrary shape
- additive noise
- nearest neighbor
- feature space
- multiscale
- neural network