• ETRI Journal
• /
• 제43권4호
• /
• pp.684-693
• /
• 2021

This paper presents a low-cost two-stage approximate multiplier for bfloat16 (brain floating-point) data processing. For cost-efficient approximate multiplication, the first stage implements Mitchell's algorithm that performs the approximate multiplication using only two adders. The second stage adopts the exact multiplication to compensate for the error from the first stage by multiplying error terms and adding its truncated result to the final output. In our design, the low-cost multiplications in both stages can reduce hardware costs significantly and provide low relative errors by compensating for the error from the first stage. We apply our approximate multiplier to the convolutional neural network (CNN) inferences, which shows small accuracy drops with well-known pre-trained models for the ImageNet database. Therefore, our design allows low-cost CNN inference systems with high test accuracy.

• 한국통신학회논문지
• /
• 제38C권11호
• /
• pp.1029-1043
• /
• 2013

CS(Compressed Sensing)는 오늘날 신호 처리 영역에서 많은 주목을 받고 있는 이론 중의 하나이다. 이 CS 분야에서 효과적인 복구 알고리즘을 설계하는 것은 가장 큰 도전적 연구 중의 하나로 인식되고 있다. 이에 따라 다양한 복구 알고리즘이 많은 문헌을 통해서 제안 되었으며 최근에 Maleki와 Donoho에 의해 제안된 AMP(Approximation Message Passing) 알고리즘은 기존에 제시된 알고리즘에 비해 간단한 구조를 가지고 있지만 좋은 성능을 보여줌으로써 상당한 주목을 받고 있다. 기존의 (BP) Belief Propagation 알고리즘은 오직 희소(Sparse) 센싱 행렬에서만 좋은 성능을 보여 준 것에 반해, AMP 알고리즘은 밀집(Dense) 센싱 행렬에 기초를 둔 Belief Propagation 알고리즘임에도 불구하고 이와 비슷한 성능을 보여준다. 본 논문은 다양한 영역에서 AMP 알고리즘이 적용되기 위하여 이에 대한 지침 및 기존의 고전적 Message Passing 알고리즘과의 관계에 대해 분석하였다. 또한 기존의 알고리즘과의 비교 분석을 통해 AMP 알고리즘의 우수성을 제시하였다.

• 한국기계가공학회지
• /
• 제20권6호
• /
• pp.33-43
• /
• 2021

A60 class bulkhead penetration piece is a fire resistance system installed on a bulkhead compartment to protect lives and to prevent flame diffusion in a fire accident on a ship and offshore plant. This study focuses on the approximate optimization of the fire resistance design of the A60 class bulkhead penetration piece using a multi-island genetic algorithm. Transient heat transfer analysis was performed to evaluate the fire resistance design of the A60 class bulkhead penetration piece. For approximate optimization, the bulkhead penetration piece length, diameter, material type, and insulation density were considered discrete design variables; moreover, temperature, cost, and productivity were considered constraint functions. The approximate optimum design problem based on the meta-model was formulated by determining the discrete design variables by minimizing the weight of the A60 class bulkhead penetration piece subject to the constraint functions. The meta-models used for the approximate optimization were the Kriging model, response surface method, and radial basis function-based neural network. The results from the approximate optimization were compared to the actual results of the analysis to determine approximate accuracy. We conclude that the radial basis function-based neural network among the meta-models used in the approximate optimization generates the most accurate optimum design results for the fire resistance design of the A60 class bulkhead penetration piece.

• Communications for Statistical Applications and Methods
• /
• 제27권2호
• /
• pp.255-268
• /
• 2020

A mixture of multivariate canonical fundamental skew t-distribution (CFUST) has been of interest in various fields. In particular, interest in the unsupervised learning society is noteworthy. However, fitting the model via EM algorithm suffers from significant processing time. The main cause is due to the calculation of many multivariate t-cdfs (cumulative distribution functions) in E-step. In this article, we provide an approximate, but fast calculation method for the in univariate fashion, which is the product of successively conditional univariate t-cdfs with Taylor's first order approximation. By replacing all multivariate t-cdfs in E-step with the proposed approximate versions, we obtain the admissible results of fitting the model, where it gives 85% reduction time for the 5 dimensional skewness case of the Australian Institution Sport data set. For this approach, discussions about rough properties, advantages and limits are also presented.

• 대한의용생체공학회:의공학회지
• /
• 제13권2호
• /
• pp.115-124
• /
• 1992

An algorithm has been proposed for the automatic detection of optimal epiand endocardial left ventricular borders from 2-D short axis echocardiogram which is degraded by noise and echo drop out. For the implementation of the algorithm, we modified Ballard's Generalized Hough Transform which can be applicable only for deterministic object border, and newly proposed Fuzzy Hough Transform method. The algorithm presented here allows detection of object whose exact shapes are unknown. The algorithm only requires an approximate model of target object based on anatomical data. To detect the approximate epicardial contour of left ventricle, Fuzzy Hough Transform was applied to the echocardiogram. The optimal epicardial contour was founded by using graph searching method which contains cost function analysis process. Using this optimal epicardial contour and average thickness imformation of left ventricular wall, the approximate endocardial line was founded, and graph searching method was also used to detect optimal endocardial contour.

• Journal of Information Processing Systems
• /
• 제6권1호
• /
• pp.67-78
• /
• 2010

In this study, a clustering algorithm that uses DCT transformed data is presented. The algorithm is a grid density-based clustering algorithm that can identify clusters of arbitrary shape. Streaming data are transformed and reconstructed as needed for clustering. Experimental results show that DCT is able to approximate a data distribution efficiently using only a small number of coefficients and preserve the clusters well. The grid based clustering algorithm works well with DCT transformed data, demonstrating the viability of DCT for data stream clustering applications.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.141.2-141
• /
• 2001

An approximate target adaptive guidance algorithm(TAG) is proposed on the basis of the assumption that angular acceleration of missile to target line-of-sight and start time for TAG can be obtained by IR seeker. The algorithm does not use any target state estimator. Instead, it avoids the problem of determining target attitude by using the observation that the missile using LOS rate guidance is nearly on the collision course in the later point of engagement. Computer simulation results show that the proposed algorithm can effectively perform target adaptive guidance.

• 한국정보과학회논문지:정보통신
• /
• 제30권3호
• /
• pp.416-423
• /
• 2003

본 논문에서는 버퍼를 공유하는 패킷 교환기의 연속 시간 큐잉 모델을 제시하고 큐 길이 확률 분포를 구하는 근사 계산 알고리즘을 제안한다. N 개의 입력 프로세스는 상호 이질적인 버스트 특성을 갖는다. 입력 프로세스는 계차-2 콕시안 분포로서 모형화하며 서버의 서비스 시간은 계차-r 얼랑 분포로서 모형화한다. 근사 알고리즘은 통합된 상태 변수를 사용하여 큐잉 시스템을 표현한다. 먼저 N개의 입력프로세스는 하나의 통합된 상태 변수로 나타내며 큐잉 시스템은 서브 시스템으로 분해하고 이것을 통합된 상태 변수로 나타낸다. 그리고 이러한 통합된 상태 변수를 사용하여 반복적인 방법에 의해서 상태 방정식의 해를 유도한다. 근사 알고리즘의 타당성은 시뮬레이션을 통해서 검증한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권12호
• /
• pp.4567-4583
• /
• 2021

This study proposes an analytical approximation algorithm based on extreme value theory (EVT) for the inverse of the power of the incomplete Gamma function. First, the Gumbel function is used to approximate the power of the incomplete Gamma function, and the corresponding inverse problem is transformed into the inversion of an exponential function. Then, using the tail equivalence theorem, the normalized coefficient of the general Weibull distribution function is employed to replace the normalized coefficient of the random variable following a Gamma distribution, and the approximate closed form solution is obtained. The effects of equation parameters on the algorithm performance are evaluated through simulation analysis under various conditions, and the performance of this algorithm is compared to those of the Newton iterative algorithm and other existing approximate analytical algorithms. The proposed algorithm exhibits good approximation performance under appropriate parameter settings. Finally, the performance of this method is evaluated by calculating the thresholds of space-time block coding and space-frequency block coding pattern recognition in multiple-input and multiple-output orthogonal frequency division multiplexing. The analytical approximation method can be applied to other related situations involving the maximum statistics of independent and identically distributed random variables following Gamma distributions.

• 한국경영과학회지
• /
• 제41권2호
• /
• pp.35-51
• /
• 2016

This paper proposes a new approximate approach to calculate the order fill rate and the probability of filling an entire customer order immediately from the shelf in a business environment under purchase dependence characterized by customer purchase patterns observed in such areas as marketing, manufacturing systems, and distribution systems. The new approximate approach divides customer orders into item orders and calculates fill rates of all order types to approximate the order fill rate. We develop a greed iterative search algorithm (GISA) based on the Gauss-Seidel method to avoid dimensionality and prevent the solution divergence for larger instances. Through the computational analysis that compares the GISA with the simulation, we demonstrate that the GISA is a dependable algorithm for deriving the stationary joint distribution of on-hand inventories in the type-K pure system. We also present some managerial insights.