• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.6
• /
• pp.567-572
• /
• 2019

In this paper, we propose an algorithm to perform convolution, pooling, and ReLU operations in CNN using binary image and binary kernel. It decomposes 256 gray-scale images into 8 bit planes and uses a binary kernel consisting of -1 and 1. The convolution operation of binary image and binary kernel is performed by addition and subtraction. Logically, it is a binary operation algorithm using the XNOR and comparator. ReLU and pooling operations are performed by using XNOR and OR logic operations, respectively. Through the experiments to verify the usefulness of the proposed algorithm, We confirm that the CNN operation can be performed by converting it to binary logic operation. It is an algorithm that can implement deep running even in a system with weak computing power. It can be applied to a variety of embedded systems such as smart phones, intelligent CCTV, IoT system, and autonomous car.

• Journal of the Korean Magnetics Society
• /
• v.5 no.3
• /
• pp.185-190
• /
• 1995

Rare earth-aluminum intermetallic compounds $RAI_{2}$ (R ; Lu, Ce, Gd) are prepared by the arc-melt method and the magnetic properties and electronic structures are investigated by magnetic susceptiptibiliy measurements using SQUID magnetometer. The magnetic suceptibiliyof $LuAl_{2}$ is weakly temperature dependent and shows a Pauli susceptibility of $10.1{\times}10^{-5}$ emu/mol, which means 3.2 states/eV/formula unit. On the other hand, the susceptibility data of $CeAl_{2}$ and $GdAl_{2}$ show a Curie-Weiss behavior for paramagnets. The magnetization data at low temperatures confirm that $CeAl_{2}$ undergoes an antiferromagnetic phase transition near 4 K whereas $GdAl_{2}$ a ferromagnetic transition at 170 K. The distinctive magnetic behaviors of $RAI_{2}$ originate from the different 4f band filling.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.520-520
• /
• 2023

제주도 동부 중산간 지역은 화산암으로 구성된 지하지질로 인해 지하수위의 변동폭이 크고 변동양상이 복잡하여 인공신경망(Artificial Neural Network, ANN) 모델 등을 활용한 지하수위의 예측이 어렵다. ANN에 적용되는 활성화함수에 따라 지하수의 예측성능은 달라질 수 있으므로 활성화함수의 비교분석 후 적절한 활성화함수의 사용이 반드시 필요하다. 본 연구에서는 5개 활성화함수(sigmoid, hyperbolic tangent(tanh), Rectified Linear Unit(ReLU), Leaky Rectified Linear Unit(Leaky ReLU), Exponential Linear Unit(ELU))를 제주도 동부 중산간지역에 위치한 2개 지하수 관정에 대해 비교분석하여 최적 활성화함수 도출을 목표로 한다. 또한 최적 활성화함수를 활용한 ANN의 적용성을 평가하기 위해 최근 널리 사용되고 있는 순환신경망 모델인 Long Short-Term Memory(LSTM) 모델과 비교분석 하였다. 그 결과, 2개 관정 중 지하수위 변동폭이 상대적으로 큰 관정은 ELU 함수, 상대적으로 작은 관정은 Leaky ReLU 함수가 지하수위 예측에 적절하였다. 예측성능이 가장 낮은 활성화함수는 sigmoid 함수로 나타나 첨두 및 최저 지하수위 예측 시 사용을 지양해야 할 것으로 판단된다. 도출된 최적 활성화함수를 사용한 ANN-ELU 모델 및 ANN-Leaky ReLU 모델을 LSTM 모델과 비교분석한 결과 대등한 지하수위 예측성능을 나타내었다. 이것은 feed-forward 방식인 ANN 모델을 사용하더라도 적절한 활성화함수를 사용하면 최신 순환신경망과 대등한 결과를 도출하여 활용 가능성이 충분히 있다는 것을 의미한다. 마지막으로 LSTM 모델은 가장 적절한 예측성능을 나타내어 다양한 인공지능 모델의 예측성능 비교를 위한 기준이 되는 참고모델로 활용 가능하다. 본 연구에서 제시한 방법은 지하수위 예측과 더불어 하천수위 예측 등 다양한 시계열예측 및 분석연구에 유용하게 사용될 수 있다.

• 한국방재학회:학술대회논문집
• /
• 2011.02a
• /
• pp.219-219
• /
• 2011

순 장파복사량은 지표면으로 입사되는 하강 장파복사량(Downward Longwave Radiation, $R_{ld}$)과 지표면에서 반사되는 상승 장파복사량(Upward Longwave Radiation, $R_{lu}$)의 차이로 정의되는데 이는 에너지 수지 및 농업기상 연구의 중요한 주제 중 하나로서 다루어져 온 순복사량의 중요한 요소이다. 일반적으로 $R_{lu}$의 경우 지표면 온도와 방사율(emissivity)를 이용하여 산출되므로 정확히 추정이 가능하나, $R_{ld}$의 경우 대기 최상층에서 관측되는 방사량과 지표면 근처의 방사량을 함께 고려해야 하므로 실측이 어렵다. $R_{ld}$는 야간 복사계(pyrgeometer)를 이용하여 직접적으로 측정할 수 있지만 관측기기 자체가 구비되어있는 관측소가 적어 매우 드물게 이용된다. 또한 단파 복사 에너지 측정 기기에 비해 비용이 많이들고 종종 관측값이 큰 오차를 가지고 있기 때문에 실무에 적용하기 힘든 단점이 있다. 따라서 기상 관측소에서 얻어지는 증기압과 온도 관측치를 물리식, 경험식 등에 적용하여 산정하게 된다. 현재는 $R_{ld}$의 추정은 관측된 방사량간의 관계를 나타내는 경험식을 기반으로 지표면 근처의 대기 온도와 습도를 이용하여 산출하는 방법이 널리 사용되고 있다. 본 연구에서는 증발산 산정 알고리즘 개발의 시발점으로써 $R_{ld}$를 먼저 구하고 $R_{lu}$를 구하였다. 신뢰성 높은 방법을 이용하여 $R_{ld}$를 구하게 되면 정확도 높은 $R_N$을 구하는 데 기여할 수 있으며, 궁극적으로 보다 정확한 증발산을 산정할 수 있게 된다. $R_{ld}$는 일반적으로 clear sky 조건 하에서의 복사 에너지 플럭스($R_{ldc}$)를 구한 후 구름의 양에 따라 보정한다. 하강 장파복사량의 경우 널리 사용되는 공식 중 하나인 Brutsaert의 공식을 사용하였다. 광릉, 해남에 위치한 플럭스 타워지점에서 실측된 기온과 실제 수증기압을 입력인자로 사용하여 지점별 $R_{ldc}$를 먼저 구하고 Moderate Resolution Imaging Spectroradiometer(MODIS) 영상자료를 이용하여 검증한 뒤 최종적으로 남한지역을 대상으로 순 장파복사량 지도를 작성하였다. 이를 위해 MODIS 07 대기 프로파일 산출물(Atmospheric Profile Product)중 기온 및 이슬점온도를 추출하여 산정식의 입력자료로서 사용하였다. 상승 장파복사량의 경우 MODIS 11 지표면 온도 산출물(Land Surface Temperature product)를 이용하여 산정하였다. 이는 남한 지역의 증발산량 추정 및 에너지 수지 연구를 위한 중요한 기본 자료로서 유용하게 사용될 수 있으리라 사료된다.

• Journal of computational fluids engineering
• /
• v.21 no.1
• /
• pp.10-18
• /
• 2016

Numerical boundary conditions are as important as the governing equations when analyzing the fluid flows numerically. An explicit boundary condition method updates the solutions at the boundaries with extrapolation from the interior of the computational domain, while the implicit boundary condition method in conjunction with an implicit time integration method solves the solutions of the entire computational domain including the boundaries simultaneously. The implicit boundary condition method, therefore, is more robust than the explicit boundary condition method. In this paper, steady compressible 2-Dimensional Navier-Stokes solver is developed. We present the implicit boundary condition method coupled with LU-SGS(Lower Upper Symmetric Gauss Seidel) method. Also, the explicit boundary condition method is implemented for comparison. The preconditioning Navier-Stokes equations are solved on unstructured meshes. The numerical computations for a number of flows show that the implicit boundary condition method can give accurate solutions.

• Journal of the Korean Geotechnical Society
• /
• v.32 no.7
• /
• pp.15-24
• /
• 2016

To investigate the unsaturated characteristics of the Boeun granite weathered soil, matric suction and volumetric water content were measured in both drying and wetting paths using Automated Soil-Water Characteristics Curve Apparatus. Based on the measured results, Soil-Water Characteristics Curve was estimated by van Genuchten (1980) model. The relationship between effective degree of saturation and matric suction showed the non-linear curve with S-shape and the hysteresis phenomenon occurred during drying and wetting paths. Suction Stress Characteristics Curve was estimated by the Lu and Likos (2006) model. The suction stress in drying path was constantly maintained and that in wetting path tended to increase when the effective degree of saturation was low. But the suction stress in drying path was larger than that in wetting path at the same degree of saturation when the effective degree of saturation became larger. Meanwhile, Hydraulic Conductivity Function was evaluated by the van Genuchten (1980) model which is one of the parameter estimation methods. The unsaturated hydraulic conductivity decreased with increasing the matric suction, and the decreasing velocity regarding to the matric suction in drying path was larger than that in wetting path.

• (The)Study of the Eastern Classic
• /
• no.71
• /
• pp.181-203
• /
• 2018

This study examined the structure that can explain the 'gewuzhizhi' as a medical research methodology and how to acquire knowledge from this structure. First, we derive a structure that implies the meaning of 'gewuzhizhi' shown in "Liji" and "daixue" in relation to "xueji" and "shaoyi". It is Xue- gewuzhizhi -'You(游)'. Then, we examined the purpose of the travelling Mt. Hua of Wang Lu, who was a landscape painter and medical researcher and physician and the implication of 'You' in Song Dynasty's painting theory of "Tuhua Jianwenzhi", "Linquan Gaozhi". In the course of the review, I explained the following. First, 'You' in the travelling Mt. Hua of Wang Lu and 'You' in the painting theories of Song Dynasty belongs to the structure of Xue- gewuzhizhi-'You'. Secondly, the meaning of 'gewuzhizhi' in the structure of Xue-gewuzhizhi-'You' was deepened in Song Dynasty. Third, The way in which the 'You' of the Song Dynasty's painting theory is to learn the reason of things is to observe things. Fourthly, the structure of the Xue-gewuzhizhi-'You' was practiced in the painting area in the Song Dynasty, and the field of practice was extended to the study of medicine by the landscape researcher who followed the painting theory of Song Dynasty. Fifth, through the assertion of Wang Lu's 'I take Mt. Hua as my teacher[師]', Wang Lu finds that the source of the recognition lies in nature(Mt. Hua). Through the above, 'gewuzhizhi' shown in "Liji" and "dai xue" is involved in the theory of cultivation and epistemology, and the structure of xue-gewuzhizhi- 'You' involved in epistemology and the method of recognition is found in observation. Through these investigations, I understood logically 'gewuzhizhi' as a method of medical research.

• The Journal of Korean Society of Virology
• /
• v.26 no.2
• /
• pp.235-244
• /
• 1996

We selected the adequate cell line to be used for propagation and plaquing of R. tsutsugamushi in laboratory and identified R. tsutsugamushi serotype cultured in LuMA cells by nested PCR. As in this study, we concluded that. 1. LuMA cell was suitable for the study of the biology of rickettsiae-host cell interaction. 2. The plaque-forming unit (PFU) per ml of R. tsutsugamushi Karp strain propagated in embryonated egg yolk sacs was $10^{8.8}$ and the PFU/ml of Gilliam strain was $10^{7.1}$. 3. The rate and extent of cytopathic changes depended on the PFU titer of R. tsutsugamushi. 4. PCR with nested primer pairs was useful for identification of R. tsutsugamushi serotype cultured in human embryonic lung cells.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.63-68
• /
• 2022

Autonomous driving technology has recently made great progress with the introduction of deep reinforcement learning. In order to effectively use deep reinforcement learning, it is important to select the appropriate activation function. In the meantime, many activation functions have been presented, but they show different performance depending on the environment to be applied. This paper compares and evaluates the performance of 12 activation functions to see which activation functions are effective when using reinforcement learning to learn autonomous driving on highways. To this end, a performance evaluation method was presented and the average reward value of each activation function was compared. As a result, when using GELU, the highest average reward could be obtained, and SiLU showed the lowest performance. The average reward difference between the two activation functions was 20%.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.5 no.3
• /
• pp.178-182
• /
• 1981

쉘이 복합재로 제작되었거나 보강재를 사용하였을때에는 쉘요소의 역학적 특성은 방향성을 가지게 된다. 이 논문에서는 선형관계식을 사용하여 동적 표면가동을 받는 이방성원통쉘의 안정성을 고려하여 다루었다. 재료의 이방성을 고려하여 3방향 즉, 축방향, 둘레방향 및 반경향의 실성항이 포함된 운동방장실을 유도하였다. 동적 안전성의 해석은 Bolotin 의 방법을 따랐으며 근사해를 행렬식행으로 제시하였다. 수직열로 외압을 받는 원통쉘의 좌단을 다루었으며 본 이론과 고전이론 사이의 차이점을 검토하였다.