• 대한임베디드공학회논문지
• /
• 제17권5호
• /
• pp.297-307
• /
• 2022

The simple procedural segment selection algorithm commonly used in Dynamic Adaptive Streaming over HTTP (DASH) reveals severe weakness to provide high-quality streaming services in the integrated mobile networks of various wired and wireless links. A major issue could be how to properly cope with dynamically changing underlying network conditions. The key to meet it should be to make the segment selection algorithm much more adaptive to fluctuation of network traffics. This paper presents a system architecture that replaces the existing procedural segment selection algorithm with a deep reinforcement learning algorithm based on the Asynchronous Advantage Actor-Critic (A3C). The distributed A3C-based deep learning server is designed and implemented to allow multiple clients in different network conditions to stream videos simultaneously, collect learning data quickly, and learn asynchronously, resulting in greatly improved learning speed as the number of video clients increases. The performance analysis shows that the proposed algorithm outperforms both the conventional DASH algorithm and the Deep Q-Network algorithm in terms of the user's quality of experience and the speed of deep learning.

• 한국수산과학회지
• /
• 제54권6호
• /
• pp.965-973
• /
• 2021

The behavior of angelfish Pterophyllum scalare exposed to low and high temperatures was monitored by video tracking, and information such as the initial speed, changes in speed, and locations of the fish in the tank were analyzed. The water temperature was raised from 26℃ to 36℃ or lowered from 26℃ to 16℃ for 4 h. The control group was maintained at 26℃ for 8 h. The experiment was repeated five times for each group. Machine learning analysis comprising a long short-term memory model was used to train and test the behavioral data (80 s) after pre-processing. Results showed that when the water temperature changed to 36℃ or 16℃, the average speed, changes in speed and fractal dimension value were significantly lower than those in the control group. Machine learning analysis revealed that the accuracy of 80-s video footage data was 87.4%. The machine learning used in this study could distinguish between the optimal temperature group and changing temperature groups with specificity and sensitivity percentages of 86.9% and 87.4%, respectively. Therefore, video tracking technology can be used to effectively analyze fish behavior. In addition, it can be used as an early warning system for fish health in aquariums and fish farms.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1997년도 추계학술대회 논문집 학회본부
• /
• pp.505-508
• /
• 1997

In this paper, an auto-tuning method for fuzzy controller based on the neural network is presented. The backpropagated error of neural emulator offers the path which reforms the fuzzy controller's membership functions and fuzzy rule, and used for speed control of induction motor. For the torque control method, an indirect vector control scheme with slip calculation is used because of its stable characteristics regardless of speed. Motor input current is regulated by a current controlled voltage source PWM inverter using space voltage vector technique. Also, the scheme of current control fuzzy controller is synchronous reference frame with decoupling term. DSP(TMS320C31) is used to achieve the high speed calculation of the space voltage vector PWM and to build the self-learning fuzz. control algorithm. An IPM is used to simplify hardware design.

• 전기전자학회논문지
• /
• 제23권4호
• /
• pp.1337-1342
• /
• 2019

자동 한국 번호판 인식 (AKLPR)은 많은 분야에서 사용된다. 이러한 응용 분야에서 ALPR은 높은 인식률과 빠른 처리 속도가 중요하다. 최근 딥러닝의 발전으로 객체 감지 및 인식의 정확도와 속도가 향상 되고 있으며, 그 결과 딥러닝이 ALPR에 적용되고 있다. 특히 합성곱신경망(Convolutional Neural Network) 기반 객체 검출기가 ALPR에 적용되었다. 이러한 ALPR은 LP 영역을 검출하는 단계와 LP 영역의 문자를 검출 및 인식하는 단계로 구분되며, 각 단계는 별도의 CNN으로 구현된다. 본 논문에서는 단일 단계 CNN으로 ALPR을 구현하는 아키텍처를 제안한다. 제안하는 방법은 높은 인식률을 유지하면서 빠른 속도로 번호판 문자를 인식한다.

• 융합정보논문지
• /
• 제11권1호
• /
• pp.12-19
• /
• 2021

원활한 작전 수행을 통한 국방력의 강화를 위해 전술네트워크의 기능은 필수적이다. 전시 상황에서 다양한 전술, 전략은 수많은 정보들을 근거로 한다. 이를 위해 정찰기를 비롯한 다양한 정보 수집 장치 및 자원들이 방대한 양의 정보 수집을 위해 사용되고, 이들 대다수는 전술네트워크를 통해 정보를 전달한다. 채널의 사용 여부를 판단하여 상황에 따라 경쟁 기반으로 채널에 접속을 하는 국방전술네트워크 환경에서, 매우 높은 이동성을 갖는 정찰기 등 고속 이동 노드는 불필요한 채널 점유로 인하여 잠재적인 성능 열화 문제가 발생할 수 있다. 본 논문에서는 채널 예약 시점을 정하는 경쟁 윈도우(Contention Window)의 크기를 경험적으로 학습시켜 네트워크 처리량을 증가시키는 Learning-Backoff 방식의 무전 채널 접속 방법을 제안한다. 제안하는 방법은 고속 이동 노드의 수가 많아짐에 따라 더욱 좋은 성능을 보이고 있으며, 정찰기 4대가 운영되는 특정 작전 시나리오에 적용하였을 경우 처리량이 최대 25% 증가한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1991년도 한국자동제어학술회의논문집(국내학술편); KOEX, Seoul; 22-24 Oct. 1991
• /
• pp.1122-1126
• /
• 1991

As microcomputers have become widespread and the high speed solenoid valves have been developed, digitally controlled hydraulic systems are used in many applications. This study deals with position control of hydraulic cylinder operated by two port 3-way high speed solenoid valve using a self-learning strategy. This was done by developing a control algorithm for the microcomputer which always automatically adjust the length of control pulse to the optimum value in accordance with the error regardless of changes in the operating condition and physical differences between components. Tests carried out in the laboratory indicate that a positional accuracy could be improved.

• 한국정보통신학회논문지
• /
• 제14권9호
• /
• pp.2031-2039
• /
• 2010

본 연구에서는 유비쿼터스 컴퓨팅 학습의 개념과 e-Learning과 u-Learning의 학습환경 및 유비쿼터스 학습환경의 설계, 그에 따른 문제점 및 해결방안을 모색하고 향후 발전방안에 대해서 고찰하였다. 최근 정보통신 및 컴퓨터의 기술발전에 따라 언제 어디서나 쉽게 원하는 학습을 할 수 있는 유비쿼터스컴퓨팅학습의 요구와 필요성이 대두되고 있다. 특히, DMB(Digital Multimedia Broadcast), WiBro, WCDMA와 같은 고속 이동 데이터 통신망의 등장과 소형의 DMB단말기, PDA, 고기능/고성능 휴대폰의 일반화는 유비쿼터스컴퓨팅 학습의 발전가능성을 더욱 촉진시킬 것이다.

• Design & Manufacturing
• /
• 제17권4호
• /
• pp.72-78
• /
• 2023

Injection molding is a process widely used in various industries because of its high production speed and ease of mass production during the plastic manufacturing process, and the product is molded by injecting molten plastic into the mold at high speed and pressure. Since process conditions such as resin and mold temperature mutually affect the process and the quality of the molded product, it is difficult to accurately predict quality through mathematical or statistical methods. Recently, studies to predict the quality of injection molded products by applying artificial neural networks, which are known to be very useful for analyzing nonlinear types of problems, are actively underway. In this study, structural optimization of neural networks was conducted by applying multi-task learning techniques according to the characteristics of the input and output parameters of the artificial neural network. A structure reflecting the characteristics of each process step was applied to the input parameters, and a structure reflecting the quality characteristics of the injection molded part was applied to the output parameters using multi-tasking learning. Building an artificial neural network to predict the three qualities (mass, diameter, height) of injection-molded product under six process conditions (melt temperature, mold temperature, injection speed, packing pressure, pacing time, cooling time) and comparing its performance with the existing neural network, we observed enhancements in prediction accuracy for mass, diameter, and height by approximately 69.38%, 24.87%, and 39.87%, respectively.

• Investigative Magnetic Resonance Imaging
• /
• 제25권1호
• /
• pp.10-22
• /
• 2021

Purpose: To overcome the difficulty in building a large data set with a high-quality in medical imaging, a concept of 'blended-transfer learning' (BTL) using a combination of both source data and target data is proposed for the target task. Materials and Methods: Source and target tasks were defined as training of the source and target networks to reconstruct cardiac CINE images from undersampled data, respectively. In transfer learning (TL), the entire neural network (NN) or some parts of the NN after conducting a source task using an open data set was adopted in the target network as the initial network to improve the learning speed and the performance of the target task. Using BTL, an NN effectively learned the target data while preserving knowledge from the source data to the maximum extent possible. The ratio of the source data to the target data was reduced stepwise from 1 in the initial stage to 0 in the final stage. Results: NN that performed BTL showed an improved performance compared to those that performed TL or standalone learning (SL). Generalization of NN was also better achieved. The learning curve was evaluated using normalized mean square error (NMSE) of reconstructed images for both target data and source data. BTL reduced the learning time by 1.25 to 100 times and provided better image quality. Its NMSE was 3% to 8% lower than with SL. Conclusion: The NN that performed the proposed BTL showed the best performance in terms of learning speed and learning curve. It also showed the highest reconstructed-image quality with the lowest NMSE for the test data set. Thus, BTL is an effective way of learning for NNs in the medical-imaging domain where both quality and quantity of data are always limited.

• 대한가정학회지
• /
• 제36권12호
• /
• pp.25-41
• /
• 1998

HALS(Hypermedia-Aided Learning System) is very useful computer networking educational system in high information society, and was developed by Kyungwon University. This system has three characteristics; 1) face to face learning, 2) ultra high speed information networing, 3) web based hypermedia courseware. So, the purpose of this study is to try the application to pre-parent eucational program(session 4 : becoming parents through the experience of pregnancy and childbirth). For this purpose it is described the definition and characteristics of HALS. Next, it is represented the model of pre-parent educational program applied to HALS and the pictures of the session 4 (initial window, help function window, etc).