In this paper, we consider the problem of active learning, and give a neces
sary and sufficient condition of sample points for the optimal generalizati
on capability. By utilizing the properties of pseudo orthogonal bases, we c
larify the mechanism of achieving the optimal generalization capability. We
also show that the condition does not only provide the optimal generalizat
ion capability but also reduces the computational complexity and memory req
uired to calculate learning result functions. Based on the optimality condi
tion, we give design methods of optimal sample points for trigonometric pol
ynomial models. Finally, the effectiveness of the proposed active learning
method is demonstrated through computer simulations.