Support Vector Machine (SVM) Support Vector Machine (SVM)
- Problem: Optimal separation plane (decision boundary)
Optimize your Goals
- Farthest from the edge of the decision boundary
Mathematical models
Problem turned to convex optimization
Lagrange Multiplier Method--too many unknowns
KKT Transformation and Dual formula
Problem solving and neural networking
- The dual formula is a two-time programming problem, and there is a ready-made numerical method to solve
- Most Lagrange multipliers are 0 and not 0 correspond to "support vectors" (sample points that happen to be on the boundary)
- As long as the support vectors are unchanged, modifying the values of other sample points does not affect the results, and when the supporting variables change, the results will generally change.
- After solving the Lagrange multiplier, W and Bcan be introduced, and the discriminant function can be written as the following neural network style
Support Vector Machine Neural network
R and Data Analysis old notes (12) classification (support vector machines)
