• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.53 no.4
• /
• pp.265-276
• /
• 2004

Genetic Algorithm(GA), that is shown stable performance to find an optimal solution, has been used as a method of solving large-scaled optimization problems with complex constraints in various applications. Since it takes so much time to execute a long computation process for iterative evolution and adaptation. In this paper, a hardware-based adaptive GA was proposed to reduce the serious computation time of the evolutionary process and to improve the accuracy of convergence to optimal solution. The proposed GA, based on steady-state model among continuos generation model, performs an adaptive mutation process with consideration of the evolution flow and the population diversity. The drawback of the GA, premature convergence, was solved by the proposed adaptation. The Performance improvement of convergence accuracy for some kinds of problem and condition reached to 5-100% with equivalent convergence speed to high-speed algorithm. The proposed adaptive GAP(Genetic Algorithm Processor) was implemented on FPGA device Xilinx XCV2000E of EHW board for face recognition.

• Electronics and Telecommunications Trends
• /
• v.33 no.4
• /
• pp.1-13
• /
• 2018

Deep learning (DL) models have been widely applied to AI applications such image recognition and language translation with big data. Recently, DL models have becomes larger and more complicated, and have merged together. For the accelerated training of a large-scale deep learning model, model parallelism that partitions the model parameters for non-shared parallel access and updates across multiple machines was provided by a few distributed deep learning frameworks. Model parallelism as a training acceleration method, however, is not as commonly used as data parallelism owing to the difficulty of efficient model parallelism. This paper provides a comprehensive survey of the state of the art in model parallelism by comparing the implementation technologies in several deep learning frameworks that support model parallelism, and suggests a future research directions for improving model parallelism technology.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.27 no.5
• /
• pp.775-783
• /
• 1990

In this paper we implemented the fast thinning unit by modifying the pipeline architecture which was proposed by Stanley R. Sternberg. The unit is useful in preprocessing such as image representation and pattern recognition etc. This unit is composed of interface part, local memory part, address generation part, thinning processing part and control part. In thinning processing part, we shortened the thinning part which performed by means of look up table using window mapping table. Thus we improved the weakness of SAP, in which the number of delay pipeline and window pipeline are equal to image column size. Two independent memorys using tri-state buffer enable the two direction flow of address generated by address generation part. This unit avoids the complexity of architecture and has flexibility of image size by means of simple modification of logic bits.

• Journal of Digital Convergence
• /
• v.12 no.6
• /
• pp.317-323
• /
• 2014

The registration cards that are currently used to identify the people of Korea may cause secondary damage once stolen because they contain very specifically expressed fingerprint information. In order to solve this problem, in ID is required that can utilize the state-owned fingerprint DB, while while maintaining confidentiality of the identification information and satisfying the cost as well. At this point accordingly, a secure form of ID, which uses the encrypted QR code and fingerprint information, is proposed.

• Journal of the Korean Society for Precision Engineering
• /
• v.15 no.2
• /
• pp.162-169
• /
• 1998

Lacking for the skilled grinding operator possessed of the experiential knowledges in machine shop, there is the just requirement which includes the establishment of the optimal grinding conditions. Accordingly, we attemt to develope the selection system of optimal grinding conditions such as workpiece velocity, depth of cut and wheel velocity and to add the trouble shooting system by means of the neural network. Those systems are robust to the each machine error and environmental unstable state. In addition. we produce the loaming process that is progressed with additional data modified by skilled operators, and excluding is advanced to similarity of input data.

• 한국HCI학회:학술대회논문집
• /
• 2007.02a
• /
• pp.386-392
• /
• 2007

최근 들어 손 제스처를 인간-기계 인터페이스에 활용하는 연구가 많아지고 있다. 그 중에서 수화 인식은 청각장애인과 일반인 사이의 원활한 의사 소통을 하게 해 주는 인터페이스로서 중요성이 날로 더해가고 있다. 하지만 기존의 수화 인식 연구는 사용자 개개인의 수화 동작의 차이를 고려하지 않고 다수 사용자를 위한 모델을 사용하기 때문에 사용자에 따라 인식률이 낮아지게 된다. 이러한 점을 보완하기 위해 본 논문에서는 개개인의 수화 동작 특성을 반영하여 시스템이 사용자에게 적응해 가는 과정을 다루고자 한다. 특히 점진적인 사용자 적응에 있어서 가장 문제가 되는 것은 어떻게 비관측된 상태(unobserved state)의 파라미터를 수정할 것인가 하는 것이다. 이를 위해서 본 논문에서는 동작 유사도와 적응 추이에 의한 추정을 통해 비관측된 상태의 모델 파라미터를 수정한다. 실제 청각 장애인들로부터 획득한 데이터베이스를 사용하여 제안한 방법이 기존 방법에 비해 더욱 빠르게 사용자의 특성을 시스템에 반영하고 인식률을 향상시킨다는 것을 실험을 통해 보인다.

• The Journal of Korean Academic Society of Nursing Education
• /
• v.6 no.1
• /
• pp.23-35
• /
• 2000

This qualitative nursing research used a hermeneutic phenomenological approach to discovery of meaning of the nursing students' clinical experience in junior colleges. Data collecting was utilized by in- depth face-to-face interviews of 10 sophomores who were participating in their first clinical experience. The following themes of the experience emerged. 1) The first step of fear (fear, being embarrassed) 2) Rhythmical patterning of theory and practice(disappointment of nurses, recognition of the gap between the theory and practice, the conflict between medical doctors and nurses, trouble with the students from other colleges, confusion from nurses' educational and noneducational attitudes, burden of responsibility for nurses' jobs, and the tiresome nature of repeated practice) 3) Discovery of changing self, awareness of self-expansion (awareness of professional nurses' roles, formation of self-confidence, changing of one's state of mind) From this study essential theme of clinical experience of nursing student's was identified, and continuous efforts to establish better adaptation of nursing students to clinical practice are required.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.8 no.2
• /
• pp.1-8
• /
• 1998

In this paper, a method which cna detect tracking caused by the insulation deterioration of conduct wiring, is proposed. To investigate it, we analyzed the harmonics of each load current waveform and those of tracking current waveform with FFT. The computer which take experiment data is learned by neural network algorithm, which has recently been used for the load recognition. The proposed metod in our study can be applied to the development of several measuring equipments such as hotline insulation tester, cna earch tester for the detection of tracking under hot-line state, Furthermore, it can substitutes molded case circuit breaker, fuse, and so on.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2006.05a
• /
• pp.1124-1132
• /
• 2006

This paper considers a flow shop scheduling problem where a different WIP (work-in-process) state has different weight on the duration time. The objective is to minimize the sum of the weighted WIP. For the two machine flow shop case, the recognition version is unary NP-Complete. The three simple and intuitive heuristics H0, H1, and H2 are presented for the problem. For each heuristic, we find an upper bound on relative error which is tight in limit. For heuristic H2, we show that H2 dominates the other two heuristics.

• Proceedings of the KSPS conference
• /
• 2005.04a
• /
• pp.95-98
• /
• 2005

Many utterance verification (UV) algorithms have been studied to reject out-of-vocabulary (OOV) in speech recognition systems. Most of conventional confidence measures for UV algorithms are mainly based on log likelihood ratio test, but these measures take much time to evaluate the alternative hypothesis or anti-model likelihood. We propose a novel confidence measure which makes use of a momentary best scored state sequence during Viterbi search. Our approach is more efficient than conventional LRT-based algorithms because it does not need to build anti-model or to calculate the alternative hypothesis. The proposed confidence measure shows better performance in additive noise-corrupted speech as well as clean speech.