KernelMethods in Medical Imaging ぱらぱらめくる『Handbook of Biomedical Imaging Methodologies and Clinical Research』

• 機械学習を使う。サポートベクターマシン、カーネル・リッジ・回帰、カーネルPCA
• Machine learning with kernels
  • Basics
    • Classification and Regression
    • Loss function
    • Duality between features and similarity measures
    • Optimization problem over functions
  • Kernels
    • Kernels as inner products in the feature space
    • Geometric interpretation and kernel trick
    • Reproducing kernels as feature maps
    • Gaussian case
    • The overfitting problem
    • Kernels as regularizers
  • Kernelization of existing linear algorithms
    • A very simple hyperplanar classifier
    • Principal component analysis
    • kernel ridge regression and the representer theorem
  • Support vectors
    • Hyperplanar classifier in feature space and margin
    • Soft margin
    • Lagrangian approach and dual problem
    • Support vector machine
• Application to Intermodality Image Prediction
  • The MR-CT issue
  • Atlas registration vs. patches
  • Image prediction using kernel methods
  • Local learning