• 반도체디스플레이기술학회지
• /
• 제9권2호
• /
• pp.123-126
• /
• 2010

An implementation of modified stereo matching using efficient belief propagation (BP) algorithm is presented in this paper. We do recommend the use of the simple sobel, prewitt edge operator. The application of B band sobel edge operator over image demonstrates result with somewhat noisy (distinct border). When we adopt the only MRF + BP algorithm, however, borders cannot be distinguished due to that the message functions in the BP algorithm is just the mechanism which passes energy data to the only large gap of each Message functions In order to address the abovementioned disadvantageous phenomenon, we use the sobel edge operator + MRF + BP algorithm to distinguish the border that is located between the similar message data. Using edge information, the result shows that our proposed process diminishes the propagation of wrong probabilistic information. The enhanced result is due to that our proposed method effectively reduced errors incurred by ambiguous scene properties.

• Journal of information and communication convergence engineering
• /
• 제13권2호
• /
• pp.123-131
• /
• 2015

Training neural networks is a complex task with great importance in the field of supervised learning. In the training process, a set of input-output patterns is repeated to an artificial neural network (ANN). From those patterns weights of all the interconnections between neurons are adjusted until the specified input yields the desired output. In this paper, a new hybrid algorithm is proposed for global optimization of connection weights in an ANN. Dynamic swarms are shown to converge rapidly during the initial stages of a global search, but around the global optimum, the search process becomes very slow. In contrast, the gradient descent method can achieve faster convergence speed around the global optimum, and at the same time, the convergence accuracy can be relatively high. Therefore, the proposed hybrid algorithm combines the dynamic particle swarm optimization (DPSO) algorithm with the backpropagation (BP) algorithm, also referred to as the DPSO-BP algorithm, to train the weights of an ANN. In this paper, we intend to show the superiority (time performance and quality of solution) of the proposed hybrid algorithm (DPSO-BP) over other more standard algorithms in neural network training. The algorithms are compared using two different datasets, and the results are simulated.

• 한국정보처리학회논문지
• /
• 제7권2호
• /
• pp.552-557
• /
• 2000

본 논문에서는 신경망의 초기 파라미터(가중치, 바이어스) 값을 최적화 시키는 GA-BP(Genetic Algorithm-Backpropagation Network) 혼합 알고리즘을 이용하여 얼굴을 인식하는 방법을 제안하였다. 입력 영상의 각 픽셀들을 신경망의 입력으로 사용하고 고정 소수점 실수값으로 이루어진 신경망의 초기 파리미터 값은 유전자 알고리즘의 개체로 사용하기 위해 비트 스트링으로 변환한다. 신경망의 오차가 최소가 되는 값을 적합도로 정의한 뒤 새롭게 정의된 적응적 재학습 연산자를 이용하여 이를 평가해 최적의 진환된 신경망을 구성한 뒤 얼굴을 인식하는 실험을 하였다. 실험 결과 학습 수렴 속도의 비교에서는 오류 역전과 알고리즘 단독으로 실행한 수렴 속도보다 제안된 알고리즘의 수렴 속도가 향상된 결과를 보였고 인식률에서 오류 역전과 알고리즘 단독으로 실행한 방법보다 2.9% 향상된 것으로 나타났다.

• Journal of Electrical Engineering and Technology
• /
• 제10권1호
• /
• pp.41-46
• /
• 2015

It is very important to make accurate forecast of wind power because of its indispensable requirement for power system stable operation. The research is to predict wind power by chaos and BP artificial neural networks (CBPANNs) method based on genetic algorithm, and to evaluate feasibility of the method of predicting wind power. A description of the method is performed. Firstly, a calculation of the largest Lyapunov exponent of the time series of wind power and a judgment of whether wind power has chaotic behavior are made. Secondly, phase space of the time series is reconstructed. Finally, the prediction model is constructed based on the best embedding dimension and best delay time to approximate the uncertain function by which the wind power is forecasted. And then an optimization of the weights and thresholds of the model is conducted by genetic algorithm (GA). And a simulation of the method and an evaluation of its effectiveness are performed. The results show that the proposed method has more accuracy than that of BP artificial neural networks (BP-ANNs).

• 한국정보전자통신기술학회논문지
• /
• 제7권4호
• /
• pp.223-228
• /
• 2014

LT 부호에 대해 BP 복호 알고리즘은 가장 빠른 복호 방법 중 하나로 알려져 있다. 그러나 BP 알고리즘은 대부분의 LT 부호를 복호하는데 있어서 많은 오버헤드를 요구하며 특히 짧은 길이의 LT 부호에 대해서는 과다한 오버헤드가 소요된다. 본 논문에서는 오버헤드를 줄이기 위해 1-차수의 패킷을 탐색할 수 있는 방법을 제시하고 이를 이용한 개선된 BP 복호 알고리즘을 제안하였다. 제안된 복호 알고리즘은 기존의 BP 알고리즘에 비해 같은 복호 복잡도를 유지하면서도 더 적은 오버헤드를 가짐을 알 수 있었다.

• KIEE International Transactions on Electrophysics and Applications
• /
• 제5C권3호
• /
• pp.138-142
• /
• 2005

In this paper, we compared recognition rates between NN(neural networks) and clustering method as a scheme of off-line PD(partial discharge) diagnosis which occurs at the stator coil of traction motor. To acquire PD data, three defective models are made. PD data for classification were acquired from PD detector. And then statistical distributions are calculated to classify model discharge sources. These statistical distributions were applied as input data of two classification tools, BP(Back propagation algorithm) and ANFIS(adaptive network based fuzzy inference system) pre-processed FCM(fuzzy c-means) clustering method. So, classification rate of BP were somewhat higher than ANFIS. But other items of ANFIS were better than BP; learning time, parameter number, simplicity of algorithm.

• Journal of Communications and Networks
• /
• 제15권4호
• /
• pp.411-421
• /
• 2013

Digital fountain codes are record-breaking codes for erasure channels. They have many potential applications in both wired and wireless communications. Most existing digital fountain codes operate over binary fields using an iterative belief-propagation (BP) decoding algorithm. In this paper, we propose a new iterative decoding algorithm for both binary and nonbinary fields. The basic form of our proposed algorithm considers both degree-1 and degree-2 check nodes (instead of only degree-1 check nodes as in the original BP decoding scheme), and has linear complexity. Extensive simulation demonstrates that it outperforms the original BP decoding scheme, especially for a small number of source packets. The enhanced form of the proposed algorithm combines the basic form of the algorithm and a guess-based algorithm to further improve the decoding performance. Simulation results demonstrate that it can provide better decoding performance than the guess-based algorithm with fewer guesses, and can achieve decoding performance close to that of the maximum likelihood decoder at a much lower decoding complexity. Last, we show that our nonbinary scheme has the potential to outperform the binary scheme when choosing suitable degree distributions, and furthermore it is insensitive to the size of the Galois field.

• 대한전자공학회논문지SP
• /
• 제48권2호
• /
• pp.88-95
• /
• 2011

스테레오 매칭은 두 영상의 차이를 이용하여 거리를 추정하는 연구 분야로 성능 개선과 함께 처리속도 향상을 위한 연구가 계속되고 있다. 본 논문에서는 계층적 Belief Propagation(BP) 알고리즘을 개선하여 기존의 BP에서의 수렴구간을 메시지 맵으로 만들고 이를 이용하여 처리속도를 향상시키는 Plane-converging BP 알고리즘을 제안한다. 또한 GPU 아키텍쳐인 Nvidia의 CUDA를 이용하여 다수의 계산을 병렬화 하고 이를 동시에 처리하여 실시간 어플리케이션에 적합한 스테레오 매칭 기법을 개발하였다. Plane-converging BP 알고리즘은 기존의 계층적 BP 알고리즘과 유사한 에러율을 가지면서 약 2.7배의 속도 향상을 이루었다.

• 대한임베디드공학회논문지
• /
• 제4권4호
• /
• pp.195-200
• /
• 2009

As 4G mobile communication systems require high transmission rates with reliability, the need for efficient error correcting code is increasing. In this paper, a novel LDPC (Low Density Parity Check) decoder is introduced. The LDPC code is one of the most popular error correcting codes. In order to improve performance of the LDPC decoder, we use SNR (Signal-to-Noise Ratio) estimation results to adjust coefficients of modified UMP-BP (Uniformly Most Probable Belief Propagation) algorithm which is one of widely-used LDPC decoding algorithms. An advantage of Modified UMP-BP is that it is amenable to implement in hardware. We generate the optimal values by simulation for various SNRs and coefficients, and the values are stored in a look-up table. The proposed decoder decides coefficients of the modified UMP-BP based on SNR information. The simulation results show that the BER (Bit Error Rate) performance of the proposed LDPC decoder is better than an LDPC decoder using a conventional modified UMP-BP.

• 한국정보처리학회논문지
• /
• 제4권3호
• /
• pp.747-753
• /
• 1997

함수근사는 과학과 공학분야에서 광범위하게 응용된다. 다층 전방향 신경망은 비선형 함수근사를 위한 도구로서 제안되어져 왔으며, 다층 신경망을 학습시키기 위한 학습 알고리즘으로 역전파 알고리즘이 널리 이용되어져 왔다. 그러나 이상치(outlier) 를 포함한 학습자료가 존재할 때에는 학습되는 함수는 존재하는 모든 자료 사이를 보간 하므로 이상치가 있는 자료의 위치까지도 보간 하여, 원하지 않은 구조를 파악하게 된다. 따라서 이상치의 영향을 최소화 시키기 위해 본 논문에서는 로버스트 에너지 함수를 유도하여 개량된 로버스트 역전파 알고리즘을 제안한다.