• 인터넷정보학회논문지
• /
• 제22권2호
• /
• pp.59-68
• /
• 2021

최근 급속도로 성장하고 있는 인공지능 기술이 자율운항선박과 같은 해상 환경에서도 적용되기 시작하면서 디지털 영상에 특화된 CNN 기반의 모델을 적용하는 관련 연구가 활발히 진행되고 있다. 이러한 해상 서비스의 경우 인적 과실을 줄이기 위해 충돌 위험이 있는 부유물을 감지하거나 선박 내부의 화재 등 여러 가지 기술이 접목되기에 실시간 처리가 매우 중요하다. 그러나 기능이 추가될수록 프로세서의 제품 가격이 증가하는 문제가 존재해 소형 선박의 선주들에게는 비용적인 측면에서 부담이 된다. 또한 대형 선박의 경우 자율운항선박의 시스템을 감안할 때, 연산 속도의 성능 향상을 위해 복잡도가 높은 딥러닝 모델의 성능을 개선하는 방법이 필요하다. 따라서 본 논문에서는 딥러닝 모델에 경량화 기법을 적용해 정확도를 유지하면서 고속으로 처리할 수 있는 방법에 대해 제안한다. 먼저 해상 부유물 검출에 적합한 영상 전처리를 진행하여 효율적으로 CNN 기반 신경망 모델 입력에 영상 데이터가 전달될 수 있도록 하였다. 또한, 신경망 모델의 알고리즘 경량화 기법 중 하나인 학습 후 파라미터 양자화 기법을 적용하여 모델의 메모리 용량을 줄이면서 추론 부분의 처리 속도를 증가시켰다. 양자화 기법이 적용된 모델을 저전력 임베디드 보드에 적용시켜 정확도와 처리 속도를 사용하는 임베디드 성능을 고려하여 설계하는 방법을 제안한다. 제안하는 방법 중 정확도 손실이 제일 최소화되는 모델을 활용해 저전력 임베디드 보드에 비교하여 기존보다 최대 4~5배 처리 속도를 개선할 수 있었다.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권8호
• /
• pp.578-584
• /
• 2004

The CSVM(Current Support Vector Machine) that is a digital architecture performing all phases of recognition process including kernel computing, learning, and recall of SVM(Support Vector Machine) on a chip is proposed. Concurrent operation by parallel architecture of elements generates high speed and throughput. The classification problems of bio data having high dimension are solved fast and easily using the CSVM. Quadratic programming in original SVM learning algorithm is not suitable for hardware implementation, due to its complexity and large memory consumption. Hardware-friendly SVM learning algorithms, kernel adatron and kernel perceptron, are embedded on a chip. Experiments on fixed-point algorithm having quantization error are performed and their results are compared with floating-point algorithm. CSVM implemented on FPGA chip generates fast and accurate results on high dimensional cancer data.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2006년도 춘계학술대회 논문집
• /
• pp.309-314
• /
• 2006

Interior permanent magnet synchronous motor(IPMSM) has become a popular choice in electric vehicle applications, due to their excellent power to weight ratio. This paper proposes maximum torque control of IPMSM drive using adaptive learning fuzzy neural network and artificial neural network. This control method is applicable over the entire speed range which considered the limits of the inverter's current md voltage rated value. For each control mode, a condition that determines the optimal d-axis current $i_d$ for maximum torque operation is derived. This paper considers the design and implementation of novel technique of high performance speed control for IPMSM using adaptive teaming fuzzy neural network and artificial neural network. The hybrid combination of neural network and fuzzy control will produce a powerful representation flexibility and numerical processing capability. Also, this paper proposes speed control of IPMSM using adaptive teaming fuzzy neural network and estimation of speed using artificial neural network. The back propagation neural network technique is used to provide a real time adaptive estimation of the motor speed. The proposed control algorithm is applied to IPMSM drive system controlled adaptive teaming fuzzy neural network and artificial neural network, the operating characteristics controlled by maximum torque control are examined in detail. Also, this paper proposes the analysis results to verify the effectiveness of the adaptive teaming fuzzy neural network and artificial neural network.

• 전력전자학회논문지
• /
• 제3권3호
• /
• pp.173-183
• /
• 1998

본 논문은 신경회로망에 의한 퍼지제어기의 소속함수를 자동동조하는 방법을 제시하였다. 신경회로망 에뮬레이터는 퍼지제어기의 소속함수와 퍼지규칙을 재구성하는 경로를 제공하며, 재구성된 퍼지제어기는 유도전동기의 속도제어를 위해 사용한다. 따라서, 연산 시간과 시스템 성능의 관점에서 제안된 방법은 전동기 상수가 변동될 시에도 기존의 제어 방식보다 우수하다. 공간전압벡터 PWM 발생을 위한 고속연산을 수행하고 자기학습형 퍼지제어기 알고리즘을 구현하기 위해서 32비트 마이크로프로세서인 DSP(TMS320C31)을 사용하였다. 컴퓨터 시뮬레이션과 실험 결과를 통하여, 제안된 방식이 PI 제어기나 기존의 퍼지제어기보다 향상된 제어 성능을 보일 수 있음을 확인하였다.

• 대한조선학회논문집
• /
• 제58권2호
• /
• pp.105-111
• /
• 2021

The main source of underwater radiated noise of ships is cavitation generated by propeller blades. After the Cavitation Inception Speed (CIS), noise level at all frequencies increases severely. In determining the CIS, it is based on the results observed with the naked eye during the model test, however accuracy and consistency of CIS values are becoming practical issues. This study was carried out with the aim of developing a technology that can automatically recognize cavitation images using deep learning technique based on a Convolutional Neural Network (CNN). Model tests on a three-dimensional hydrofoil were conducted at a cavitation tunnel, and tip vortex cavitation was strictly observed using a high-speed camera to obtain analysis data. The results show that this technique can be used to quantitatively evaluate not only the CIS, but also the amount and rate of cavitation from recorded images.

• 전자통신동향분석
• /
• 제38권5호
• /
• pp.34-50
• /
• 2023

Quantum error correction is a key technology for achieving fault-tolerant quantum computation. Finding the best decoding solution to a single error syndrome pattern counteracting multiple errors is an NP-hard problem. Consequently, error decoding is one of the most expensive processes to protect the information in a logical qubit. Recent research on quantum error decoding has been focused on developing conventional and neural-network-based decoding algorithms to satisfy accuracy, speed, and scalability requirements. Although conventional decoding methods have notably improved accuracy in short codes, they face many challenges regarding speed and scalability in long codes. To overcome such problems, machine learning has been extensively applied to neural-network-based error decoding with meaningful results. Nevertheless, when using neural-network-based decoders alone, the learning cost grows exponentially with the code size. To prevent this problem, hierarchical error decoding has been devised by combining conventional and neural-network-based decoders. In addition, research on quantum error decoding is aimed at reducing the spacetime decoding cost and solving the backlog problem caused by decoding delays when using hardware-implemented decoders in cryogenic environments. We review the latest research trends in decoders for quantum error correction with high accuracy, neural-network-based quantum error decoders with high speed and scalability, and hardware-based quantum error decoders implemented in real qubit operating environments.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1996년도 하계학술대회 논문집 B
• /
• pp.808-810
• /
• 1996

In this paper, neural network, which has learning capability, is used for fault type classification and fault section estimation for high speed relaying. The potential of the neural network approach is demonstrated by simulation using ATP. The instantaneous values of voltages and currents are used the inputs of neural networks. This approach determines the fault section directly. In this paper, back-propagation network(BPN) is used for fault type classification and fault section estimation and can use for high speed relaying because it determines fault section within a few msec.

• 대한임베디드공학회논문지
• /
• 제14권5호
• /
• pp.277-286
• /
• 2019

In this paper, we propose an efficient hyperplane generation technique to classify human activity from combination of events and sequence information obtained from multiple-event sensors. By generating hyperplane efficiently, our machine learning algorithm classify with less memory and run time than the LSVM (Linear Support Vector Machine) for embedded system. Because the fact that light weight and high speed algorithm is one of the most critical issue in the IoT, the study can be applied to smart home to predict human activity and provide related services. Our approach is based on reducing numbers of hyperplanes and utilizing robust string comparing algorithm. The proposed method results in reduction of memory consumption compared to the conventional ML (Machine Learning) algorithms; 252 times to LSVM and 34,033 times to LSTM (Long Short-Term Memory), although accuracy is decreased slightly. Thus our method showed outstanding performance on accuracy per hyperplane; 240 times to LSVM and 30,520 times to LSTM. The binarized image is then divided into groups, where each groups are converted to binary number, in order to reduce the number of comparison done in runtime process. The binary numbers are then converted to string. The test data is evaluated by converting to string and measuring similarity between hyperplanes using Levenshtein algorithm, which is a robust dynamic string comparing algorithm. This technique reduces runtime and enables the proposed algorithm to become 27% faster than LSVM, and 90% faster than LSTM.

• International Journal of Internet, Broadcasting and Communication
• /
• 제13권2호
• /
• pp.208-217
• /
• 2021

3D-CNN is one of the deep learning techniques for learning time series data. However, these three-dimensional learning can generate many parameters, requiring high performance or having a significant impact on learning speed. We will use these 3D-CNNs to learn hand gesture and find the parameters that showed the highest accuracy, and then analyze how the accuracy of 3D-CNN varies through input data changes without any structural changes in 3D-CNN. First, choose the interval of the input data. This adjusts the ratio of the stop interval to the gesture interval. Secondly, the corresponding interframe mean value is obtained by measuring and normalizing the similarity of images through interclass 2D cross correlation analysis. This experiment demonstrates that changes in input data affect learning accuracy without structural changes in 3D-CNN. In this paper, we proposed two methods for changing input data. Experimental results show that input data can affect the accuracy of the model.

• 한국정보시스템학회지:정보시스템연구
• /
• 제18권2호
• /
• pp.111-126
• /
• 2009

A high-speed Internet has brought a rapid spread of Web Based Distance Learning(WBDL). Even though the WBDL was considered a new methodology to overcome the limitation of a traditional education, it evolves not as alternatives but as strategic augmenting tools for a traditional face-to-face education. The WBDL systems accommodate diverse services such as e-Learning, e-Mentoring, and Blended Learning in order to give satisfactions to learners and increase the learning effectiveness. This study suggested the WBDL system's and learner's characteristics as two major affecting factors, in which two independent variables were respectively selected. A mediating effect of learning motivation between the independent variables and learning achievement was empirically tested. The interactions between the WBDL sysrem and learners were also tested on the view points of the moderating effects between the learning motivation and the learning achievement. The results showed that the mediating effects of learning motivation and the moderating effects of the system interactions were statistically significant.