How could I solve this problem?

cvx_begin quiet
variable alpha_SS(N_S) nonnegative
variable alpha_ST(N_S) nonnegative
variable alpha_TT(N_T) nonnegative
variable kc_S(N_S) nonnegative
variable kc_T(N_T) nonnegative
variable b(2)
minimize( norm(alpha_TT,1 ) + norm(alpha_SS,1) + norm(alpha_ST,1 ) + C_T * norm(kc_T,1) + C_S * norm(kc_S,1))
subject to
0 <= labels_train_T.*( ( ( labels_train_T.*alpha_TT )’*KernelProcess(kernel, inst_train_T, inst_train_T) )’ + ( ( labels_S.alpha_ST)'KernelProcess(kernel, inst_S, inst_train_T) )’ + b(1) )+ kc_T - 1;
0 <= labels_S.
( ( ( labels_S.
( alpha_SS + alpha_ST ) )’*KernelProcess(kernel, inst_S, inst_S) )’ + b(2) )+ kc_S - 1

        dot(alpha_SS,alpha_ST) == 0    %Disciplined convex programming error:
                                                            % Invalid quadratic form(s): not a square.
cvx_end