Module 6: Dimensionality Reduction
  Lecture 32: Embedding
 

                                            

 

 

Lipschitz embeddings
  • Embedding of database with metric distance onto -dimensional vector space
  • Choose subsets of
  • Each subset is called a reference set
  • Define distance of object to set as
  • Feature vector
  • In earlier example
 
  • and
  • Contractive under (after scaling)
  • Says nothing about distortion