Hard-label Manifolds: Unexpected Advantages of Query Efficiency for Finding On-manifold Adversarial Examples.
Washington GarciaPin-Yu ChenSomesh JhaScott ClouseKevin R. B. ButlerPublished in: CoRR (2021)
Keyphrases
- manifold learning
- low dimensional
- finding similar
- euclidean space
- database
- query processing
- riemannian manifolds
- manifold ranking
- nonlinear dimensionality reduction
- response time
- manifold structure
- user queries
- active learner
- manifold embedding
- locally linear
- high dimensional
- feature space
- lie group
- data structure
- tangent space
- laplacian eigenmaps
- query expansion
- geodesic distance
- geometric structure
- data points
- dimension reduction
- semi supervised learning
- data sources
- active learning
- high dimensional data
- label noise
- range queries