How to Teach Support Vector Machine to Learn Vector Outputs
Lecture slides:
- The “Classical” Support Vector Machine(SVM)
- Reinterpretation of the normal vector w
- Affine transformation = Linear transformation
- Primal problem
- Dual problem
- Primal problem
- Solution
- Prediction
- Prediction when the labels are implicit
- Representation of multiclass output
- Vertices of hyper-tetrahedron
- Embedding Hierarchy
- Methods
- WIPO-alpha dataset
- Computational times
- Multiview learning
- Primal problem
- One-class SVM interpretation
- Spherical embedding
- Embedding Hierarchy
- Similarities in a hierarchy
- WIPO-alpha dataset
- Primal problem
- To get rid of occurrences of explicit labels ...
- Prediction when the labels are implicit
Author: Sandor Szedmak, University of Southampton