• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.05c
• /
• pp.336-341
• /
• 2002

본 연구에서는 엔트로피 최적화를 위한 목적함수의 근을 구하기 위해 단순히 함수 값만을 이용하여 계산을 근사화한 할선법에 기초한 고정점 알고리즘의 독립성분분석 기법을 제안하였다. 이렇게 하면 기존의 뉴우턴법에 기초한 고정점 알고리즘에서 요구되는 복잡한 도함수의 계산과정을 간략화 할 수 있어 더 우수한 학습성능의 독립성분분석이 가능하다. 제안된 학습알고리즘의 독립성분분석 기법을 512$\times$512의 픽셀을 가지는 10개의 영상을 대상으로 임의의 혼합행렬에 따라 발생되는 혼합영상들을 실험하였다. 실험결과, 기존의 뉴우턴법에 기초한 고정점 알고리즘의 분석기법보다 빠른 학습속도와 개선된 분리성능이 있음을 확인하였다. 특히 기존의 알고리즘에서 임의로 설정되는 초기값에 덜 의존하는 학습성능이 있음도 확인할 수 있었다.

• Annual Conference of KIPS
• /
• 2015.04a
• /
• pp.875-877
• /
• 2015

본 논문은 위치기반 동역학 방법에 사용되는 3D물체를 근사화하기 위한 방향성 파티클 배치방법을 소개한다. 물체 매쉬를 분할하고 각 분할된 내부에 파티클을 채우는 배치 방법을 사용하면 물체의 표면에 여러 개의 파티클을 무작위로 배차하는 기존 방법보다 방향성 파티클의 개수를 줄일 수 있다. 시뮬레이션 실험을 통해 본 논문에서 제안한 방법으로 배치한 파티클 구조가 기존 방법으로 배치한 파티클 구조에 비해 보다 빠르게 시뮬레이션을 계산할 수 있다는 것을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.280-282
• /
• 2020

최근 사물인터넷(IoT), 자율주행과 같이 기계 간의 통신이 요구되는 서비스가 늘어감에 따라, 기계 임무 수행에 최적화된 데이터의 생성 및 압축에 대한 필요성이 증가하고 있다. 또한, 사물인터넷과 인공지능(AI)이 접목된 기술이 주목을 받으면서 딥러닝 모델에서 추출되는 특징(feature)을 디바이스에서 클라우드로 전송하는 방안에 관한 연구가 진행되고 있으며, 국제 표준화 기구인 MPEG에서는 '기계를 위한 부호화(Video Coding for Machine: VCM)'에 대한 표준 기술 개발을 진행 중이다. 딥러닝으로 특징을 추출하는 가장 대표적인 방법으로는 합성곱 신경망(Convolutional Neural Network: CNN)이 있으며, 오토인코더는 입력층과 출력층의 구조를 동일하게 하여 출력을 가능한 한 입력에 근사시키고 은닉층을 입력층보다 작게 구성하여 차원을 축소함으로써 데이터를 압축하는 딥러닝 기반 이미지 압축 방식이다. 이에 본 논문에서는 이러한 오토인코더의 성질을 이용하여 CNN 기반의 이미지 분류 네트워크의 합성곱 신경망으로부터 추출된 feature에 오토인코더를 적용하여 압축하는 방안을 제안한다.

• Journal of KIISE:Computer Systems and Theory
• /
• v.29 no.7
• /
• pp.377-383
• /
• 2002

The TSP(Traveling Salesman Problem) has been known as NP-complete, there have been various studies to find the near optimal solution. The nearest neighbor heuristic is more simple than the other algorithms which are to find the optimal solution. This paper designs and implements a new distributed nearest neighbor heuristic with bounding function for the TSP using the master/slave model of PVM(Parallel Virtual Machine). Distributed genetic algorithm obtains a near optimal solution and distributed nearest neighbor heuristic finds an optimal solution for the TSP using the near optimal value obtained by distributed genetic algorithm as the initial bounding value. Especially, we get more speedup using a new genetic operator in the genetic algorithm.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.14 no.6
• /
• pp.567-574
• /
• 2003

It is important to design and evaluate the chip waveform with the minimum MAI under the bandwidth constraint in the interference-limited DS-CDMA system. In this paper, by approximation we present the analytical chip waveforms that are proposed and optimized in the reference. Their performances are compared with performances of three conventional chip waveforms: rectangular, half-sine and raised-cosine. Waveform 1 of the proposed chip waveform outperforms the conventional ones. BER and throughput performance are evaluated in the Rayleigh and Nakagami-m fading channels when DPSK modulation is used. When the required BER is 10$\^$-3/ in two fading channels, the capacity of the waveform 1 is improved about 20 % rather than raised-cosine one. When the offered traffic is 30 and the number of packet per bit(N$\sub$d/) is 14, maximum throughput of the waveform 1 is better than raised-cosine chip waveform about 18 % in two fading channels.

• Journal of Korea Multimedia Society
• /
• v.12 no.1
• /
• pp.31-40
• /
• 2009

Call centers have become the prevalent contact points between companies and their customers. By virtue of recent advances in communication technology, the number and size of call centers have grown dramatically. As a large portion of the operating costs are related to the labor costs, efficient design and workforce staffing are crucial for the economic success of call centers. Therefore it is very important to determine the adequate number of agents. In this context, the workforce staffing level can be modeled as mathematical optimization problem using queueing theory. In this paper, we deal with an approximate analysis of an N-design call center with two finite queues and two types of reneging customers by using the state decomposition method. We also represent some numerical examples and show the impact of the system parameters on the performance measures of the call center.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.34 no.8
• /
• pp.1097-1103
• /
• 2010

Injection pressure, an important factor in the filling procedure, should be minimized to enhance injection molding quality. In addition, warping deformation and weld lines, which are representative failures, should be avoided to enhance injection molding quality. To improve injection molding quality, the design procedure for an automotive fog blank cover is divided into two stages. In the first stage, we optimally obtain injection molding process variables that minimize injection pressure and warping deformation by using design of experiments, approximation and optimization techniques equipped in PIAnO (Process Integration, Automation and Optimization), a commercial PIDO (Process Integration and Design Optimization) tool. Then, we determine the thickness of the automotive fog blank cover that enables us to avoid generating weld lines. The design results we obtain in this study are found far better than those of the initial design, which demonstrates the effectiveness of our design method.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.25 no.3
• /
• pp.537-547
• /
• 2001

The paper proposes the fuzzy logic global approximate optimization strategies in optimal sizing of automotive A-pillar trim with rib structures for occupant head protection. Two different strategies referred to as evolutionary fuzzy modeling (EFM) and neuro-fuzzy modeling (NFM) are implemented in the context of global approximate optimization. EFM and NFM are based on soft computing paradigms utilizing fuzzy systems, neural networks and evolutionary computing techniques. Such approximation methods may have their promising characteristics in a case where the inherent nonlinearity in analysis model should be accommodated over the entire design space and the training data is not sufficiently provided. The objective of structural design is to determine the dimensions of rib in A-pillar, minimizing the equivalent head injury criterion HIC(d). The paper describes the head-form modeling and head impact simulation using LS-DYNA3D, and the approximation procedures including fuzzy rule generation, membership function selection and inference process for EFM and NFM, and subsequently presents their generalization capabilities in terms of number of fuzzy rules and training data.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.35 no.5
• /
• pp.575-582
• /
• 2011

A cleaning blade is an attachment installed in the toner cartridge of a laser printer for removing the residual toner from an organic photo-conductive drum. There have been many studies on the performance and life of the rubber blade. We focus on optimally designing the blade shape parameters to minimize the maximum stress of the blade while satisfying design constraints on the cleaning performance and part interference. The blade is optimally designed using a design of experiments, meta-models and an optimization algorithm implemented in PIAnO (process integration, automation, and optimization), a commercial PIDO (process integration and design optimization) tool. We integrate the CAE tools necessary for the structural analysis of the cleaning blade, automate the analysis procedure, and optimize the solution using PIAnO. We decreased the maximum stress by 32.6% in comparison with that of the initial design.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2006.05a
• /
• pp.107-108
• /
• 2006

This paper shows the design optimization of the paper feeding mechanism under dynamic behavior by using commercial codes of RecurDyn/MTT2D and RecurDyn/AutoDesign which are developed by functionBay, Inc. A virtual mockup for dynamics analysis of the paper feeding mechanism is build on RecurDyn/MTT2D and is simulated. Flexible paper is represented as a series of rigid bars connected by revolute joints and rotational spring dampers. Paper is fed by a contact and friction mechanism on rollers or guides. The slip of the paper and nip force of rollers are measured to estimate the system performance. After a simulation, these performances are automatically send to RecurDyn/AutoDesign which is a sequential approximate optimization tool based on the response surface modeling. RecurDyn/AutoDesign makes the approximate objective function and computes the optimized design points of the design variables and gives them to analysis tool. And then the simulation is repeated with the updated design variables. These processes are repeated until finding a tolerable design optimization. In this paper, a paper feeding mechanism is introduced and it is optimized with the proposed algorithms.