Artificially Intelligent Dynamic Pattern Recognition

Abstract

본 논문에서는 랜덤.피드백 루프를 가지는 디지탈학습망(Digital Learning Net)과 단층망군(Single-layer Nets)사이에 판정.피드백 루프를 두어 다범주판정시스템(Multi-category Classification System)을 구성하고, 예측되는 새로운 결과를 밝혔다.

위의 시스템은 외부입력을 내부상태에 의거하여 판독할 뿐만 아니라 판독한 결과를 안정상태에 이르도록 끊임없이 보정하는 '지능'을 가지고 동작한다.

This paper presents new results estimated with a multicategory classification system which consists of a randomly fed back digital learning net and single-layer nets, including decision feedback loops between them.

The system behaves "intelligently" not only in the sense that it clusters incoming stimuli on the basis of its internal states but also because it adjusts its decisions continuously until a stable strong decision is obtained.

This paper presents new results estimated with a multicategory classification system which consists of a randomly fed back digital learning net and single-layer nets, including decision feedback loops between them.

The system behaves "intelligently" not only in the sense that it clusters incoming stimuli on the basis of its internal states but also because it adjusts its decisions continuously until a stable strong decision is obtained.