Unsupervised Domain Adaptation Using Exemplar-SVMs with Adaptation Regularization

Algorithm 1

Domain Adaptation Exemplar Support Vector Machine.

Input:, ; parameter , , , and ;

Output: optimal and

(1) initial ;

(2) Construct kernel matrix from and based on

(6); coefficient matrix based on (7); centering matrix

; label matrix based on (11).

(3) repeat

(4)Update transformation matrix when fix

(5) Eigendecompose the optimization matrix of

and select

leading eigenvectors to construct the transformation

matrix

(6)Solve the convex optimization problem for fixed

to optimize

(7) until Convergence