Abstract
In order to learn an inverse system of a plant, the learning method which inputs the output of the plant to the learner and uses the input of the plant as the desired output signal for the learner has been used in many researches. This learning method for learning control by using a neural network was called “Direct Inverse Modeling”. Jordan regarded Direct Inverse Modeling as a purely off-line learning method and pointed out that Direct Inverse Modeling is not goal-directed. How ever, some researchers have proposed a different way of using Direct Inverse Modeling. We call on-line Direct Inverse Modeling, which is the simultaneous or alternate execution of the plant control and the inverse model learning by Direct Inverse Modeling. The learning properties of on-line Direct Inverse Modeling are completely different from that of off-line Direct Inverse Modeling. This paper shows that on-line Direct Inverse Modeling becomes goal-directed under certain conditions.