Isomap

수학노트
Pythagoras0 (토론 | 기여)님의 2020년 12월 16일 (수) 01:31 판 (→‎노트: 새 문단)
(차이) ← 이전 판 | 최신판 (차이) | 다음 판 → (차이)
둘러보기로 가기 검색하러 가기

노트

  • Isomap is distinguished by its use of the geodesic distance induced by a neighborhood graph embedded in the classical scaling.[1]
  • In a similar manner, the geodesic distance matrix in Isomap can be viewed as a kernel matrix.[1]
  • Isomap can be viewed as an extension of Multi-dimensional Scaling (MDS) or Kernel PCA.[2]
  • Isomap seeks a lower-dimensional embedding which maintains geodesic distances between all points.[2]
  • Isomap can be performed with the object Isomap .[2]
  • The algorithm can be selected by the user with the path_method keyword of Isomap .[2]
  • Isomap is viewed as a variant of metric multidimensional scaling (MDS) to model nonlinear data using its geodesic distance.[3]
  • No such type of method has been used in recent years from the Isomap perspective.[3]
  • Second section provides an overview and background work, which is followed by a brief discussion on manifold learning and Isomap.[3]
  • Isomap stands for isometric mapping.[4]
  • Isomap starts by creating a neighborhood network.[4]
  • Isomap uses the above principle to create a similarity matrix for eigenvalue decomposition.[4]
  • Classical MDS uses the euclidean distances as the similarity metric while isomap uses geodesic distances.[4]
  • We will use the neighborhood graph to achieve ISOMAP.[5]
  • kNN(k-Nearest Neighbors), the most common choice for ISOMAP, connects each data point to its k nearest neighbors.[5]
  • It uses ISOMAP and it embeds 4096 pixels and 698 face pictures into 2D space and lines them onto light direction and up-down pose.[5]

소스