초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

In this paper, we describe some shortcomings of the original adaptive resonance theory that self organize stable recognition categories regarding arbitrary sequences of input patterns in real time and study the solutions of these problems through mathematical analyses and computer simulations. As a...

In this paper, we describe some shortcomings of the original adaptive resonance theory that self organize stable recognition categories regarding arbitrary sequences of input patterns in real time and study the solutions of these problems through mathematical analyses and computer simulations. As a result we suggest an extended adaptive resonance theory(EART) neural network model. This model which is a continuous bidirectional reset mode satisfies most of the chracteristics of the general ART[l]. And it can stably recognize input patterns concerning regularly decreasing or increasing sequences. Especially, unlike the ART system which activates the reset system according to the extent of the resemblance after the learning of an input patter, this system which uses the continuous reset mode activates the reset system while learning. Thus we can reduce the total learning time. Finally, we make concrete the concept of recollection by the long long term memory (LLTM) technique.