http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
박동철 明知大學校 産業技術硏究所 1995 産業技術硏究所論文集 Vol.14 No.-
A recurrent neural network and its training algorithm are proposed in this paper. Since the proposed algorithm is based on the bilinear polynomial, it can model nonlinear systems with more parsimony than the hither order neural networks barred on the Volterra series. The proposed BiLinear Recurrent Neural Network (BLRHN) is compared with Multilayer Perception type Neural Networks(MLPNN) for time series prediction problems. The results show that the BLRNN is robust and outperforms the MLPNN in terms of Prediction accuracy.
박동철,정태균 明知大學校 産業技術硏究所 1999 産業技術硏究所論文集 Vol.18 No.-
Equalization of satellite communication using Complex-Bilinear Recurrent Neural Network(C-BLRNN) is proposed in this paper. Since the BLRNN is based on the bilinear polynomial, it has been more effectively used in modeling highly nonlinear systems with time-series characteristics than multi-layer perception type neural networks(MLPNN). The BLRNN is first expanded to its complex value version(C-BLRNN) for dealing with the complex input values. C-BLRNN is then applied to equalization of a digital satellite communication channel for M-PSK and QAM, which has severe nonlinearity with memory due to TWTA(Traveling Wave Tube Amplifier). The proposed C-BLRNN based equalizer for a channel model is compared with currently used Volterra filter Equalizer, DFE, and conventional Complex MLPNN Equalizer. The results show that the proposed C-BLRNN based equalizer gives very favorable results in both of MSE and BER criteria over Volterra filter Equalizer, DFE, and Complex MLPNN Equalizer.
경계 결정능력 향상을 위한 Query 학습에 기초한 HMM의 학습 알고리즘 연구
정지오,박동철 명지대학교 대학원 1999 대학원논문집 Vol.3 No.-
A Training algorithm of Hidden Markov Model(HMM) using Query-based learning ?? proposed and applied to the recognition of isolated digits. This paper presents a novel approach for query-based HMM learning. This algorithm uses the concept that stems from the gradient based inversion algorithm of artificial neural networks. The proposed algorithm is compared with conventional training methods on isolated digit recognition problem. The results show that the proposed method can decrease the recognition error rate up to 60% in our experiments.
정태균,박동철 명지대학교 대학원 1999 대학원논문집 Vol.3 No.-
Equalization of satellite communication using Complex-Bilinear Recurrent Neural Network(C-BLRNN) is proposed in this paper. Since the BLRNN is based on the bilinear polynomial it has been more effectively used in modeling highly nonlinear systems with time-series characteristics than multi-layer perceptron type neural networks(MLPNN). The BLRNN is first expanded to its complex value version(C-BLRNN) for dealing with the complex input values C-BLRNN is then applied to equalization of a digital satellite communication channel for M-PSK and QAM, which has severe nonlinearity with memory due to TWTA(Traveling Wave Tube Amplifier). The proposed C-BLRNN based equalizer for a channel model is compared with currently used Volterra filter Equalizer. DFE, and conventional Complex MLPNN Equalizer. The results show that the proposed C-BLRNN based equalizer gives very favorable results in both of MSE and BER criteria over Volterra filter Equalizer, DFE, and Complex MLPNN Equalizer.