A Cluster-based Approach for Improving Isotropy in Contextual Embedding Space.
Sara RajaeeMohammad Taher PilehvarPublished in: ACL/IJCNLP (2) (2021)
Keyphrases
- embedding space
- data points
- euclidean space
- clustering algorithm
- contextual information
- manifold learning
- low dimensional
- data clustering
- dimensionality reduction
- input space
- nonlinear dimensionality reduction
- graph embedding
- high dimensional
- pattern recognition
- data sets
- feature space
- knn
- nearest neighbor
- face recognition