부산대학교 도서관

The exponential rise of computational power, in embedded systems, has enhanced the overall system level computational power. This would amplify many merits of digitalization revolution in the context of ‘Industry 4.0’. The applicational needs should be captured, while designing the computational architectures. These architectures should also be aligned to take advantage of technological emerging trends. This work proposes a distributed computational architecture and possible ways of implementing ‘Edge Intelligence’. A novel neural network architecture, integrating domain knowledge (DK), is proposed for smart computation, to leverage merits of edge intelligence.