390 likes | 603 Views
Distance metric learning Vs. Fisher discriminant analysis. Different Features. Glasses vs. No Glasses. Beard vs. No Beard. Learning a Metric. Class-Equivalence Side information (Unsupervised Learning). Extension to su pervised learning. Problem statement.
E N D
Distance metric learningby Eric P. Xing, Andrew Y. Ng, Michael I. Jordan and Stuart Russell
Improving embeddings by flexible exploitation of side information.Ghodsi, A.; Wilkinson, D. F.; and Southey, F.