MULTIPLE DESCENT COST COMPETITION - RESTORABLE SELF-ORGANIZATION AND MULTIMEDIA INFORMATION-PROCESSING

Multiple descent cost competition is a composition of learning phases for minimizing a given measure of total performance, i.e., cost. If th ese phases are heterogeneous toward each other, the total learning alg orithm shows a variety of extraordinary abilities; especially in regar ds to multimedia information processing. In the first phase of descent cost learning, elements of source data are grouped. Simultaneously, a weight vector for minimal learning, (i.e., a winner), is found. Then, the winner and its partners are updated for further cost reduction. T herefore, two classes of self-organizing feature maps are generated. O ne is called a grouping feature map, which partitions the source data. The other is an ordinary weight vector feature map. The grouping feat ure map, together with the winners, retains most of the source data in formation. This feature map is able to assist in a high quality approx imation of the original data. Traditional weight vector feature maps l ack this ability. Another important capacity of the grouping feature m ap is that it can change its shape. Thus, the grouping pattern can acc ept external directions in order to metamorphose. In the text, the tot al algorithm of the multiple descent cost competition is explained fir st. In that section, image processing concepts are introduced in order to assist in the description of this algorithm. Then, a still image i s first data-compressed (DC). Next, a restored image is morphed using the grouping feature map by receiving directions given by an external intelligence. Next, an interpolation of frames is applied in order to complete animation coding (AC). Thus, multiple descent cost competitio n bridges ''DC to AC.'' Examples of multimedia processing on virtual d igital movies are given.