Isotropy in the Contextual Embedding Space: Clusters and Manifolds.
Xingyu CaiJiaji HuangYuchen BianKenneth ChurchPublished in: ICLR (2021)
Keyphrases
- embedding space
- data points
- input space
- euclidean space
- manifold learning
- low dimensional
- high dimensional
- dimensionality reduction
- geometric structure
- graph embedding
- nonlinear dimensionality reduction
- contextual information
- appearance manifolds
- clustering algorithm
- manifold structure
- geodesic distance
- high dimensional data
- euclidean distance
- smooth manifold
- feature space
- vector space
- data sets
- data clustering
- distance metric
- relative position
- input data
- kernel function
- nearest neighbor
- semi supervised
- pairwise