• 한국정보과학회논문지:데이타베이스
• /
• 제28권4호
• /
• pp.545-557
• /
• 2001

본 논문에서는 저장 데이타베이스의 정보 시스템을 정제하여 이해 가능한 정보로 전환하고 새로운 객체를 근사 추론할 수 있도록 하기 위해 러프 소속 함수 값의 개념을 도입한 계층적 근사 분류 알 고리즘을 제안한다. 제안하는 알고리즘은 근사 추론의 한 방법인 퍼지 추론 방법의 언어적 불확실성을 속 성의 퍼지 소속 함수 값으로 나타내고 조건 속성의 소속 함수 값의 합성에 의해 근사 추론하는 방법을 이용하였으며 퍼지 소속 함수 값 대신에 러프 소속 함수 값을 이용하도록 제안하였다. 이는 퍼지 소속 함 수 값을 이용하여 괴지 규칙을 생성하는 과정을 생략할 수 있는 장점이 있다. 또한 정보 시스템 내의 속 성 중에서 수치 속성에 대한 이산화 방법을 연구하고 이것 또한 러프 소속 함수 값과 정보이론의 무질서 도의 개념을 이용한 수치 속성의 이산화를 제안하였다. 제안된 알고리즘을 이용하여 패턴 분류 문제에 교 준적으로 사용되는 IRIS 데이타에 대한 실험결과96%~98% 분류율을 나타냈으며 다른 실험 데이타에서 도 기존 알고리즘과 비교하여 수치 이산화나 근사 추론 모두 우수함을 보였다.

• Journal of applied mathematics & informatics
• /
• 제4권2호
• /
• pp.323-340
• /
• 1997

In this study we prove the mesh-independence principle via Steffensen's method. This principle asserts that when Steffensen's method is applied to a nonlinear equation between some Banach spaces as well as to some finite-dimensional discretization of that equation then the behavior of th discretized process is asymptoti-cally the same as that for the original iteration. Local and semilo-cal convergencve results as well as an error analysis for Steffensen's method are also provided.

• Structural Engineering and Mechanics
• /
• 제8권5호
• /
• pp.439-452
• /
• 1999

A simple numerical scheme suitable for tracing the fracture propagation path for structures idealized by means of Hillerborg's classical cohesive crack model is presented. A direct collocation, multidomain boundary element method is adopted for the required space discretization. The algorithm proposed is necessarily iterative in nature since the crack itinerary is a priori unknown. The fracture process is assumed to be governed by a path-dependent generally nonlinear softening law. The potentialities of the method are illustrated through two examples.

• 대한수학회지
• /
• 제54권1호
• /
• pp.193-213
• /
• 2017

We investigate an efficient approximation of solution to stochastic Burgers equation driven by an additive space-time noise. We discuss existence and uniqueness of a solution through the Orstein-Uhlenbeck (OU) process. To approximate the OU process, we introduce the Karhunen-$Lo{\grave{e}}ve$ expansion, and sparse grid stochastic collocation method. About spatial discretization of Burgers equation, two separate finite element approximations are presented: the conventional Galerkin method and Galerkin-conservation method. Numerical experiments are provided to demonstrate the efficacy of schemes mentioned above.

• 한국인터넷방송통신학회논문지
• /
• 제22권5호
• /
• pp.75-86
• /
• 2022

본 연구는 딥러닝 기반 변분 오토인코더(Variational Autoencoder)를 활용하여 수치형 학습데이터 내 클래스 불균형 문제를 해결하고, 학습데이터를 증강하여 학습모델의 성능을 향상시키고자 한다. 우리는 주어진 테이블 데이터에 대하여 인위적으로 레코드 개수를 늘리기 위해 'D-VAE'을 제안한다. 제안 기법은 최적의 데이터 증강을 지원하기 위해 우선 이산화와 특징선택을 수반한 전처리 과정을 수행한다. 이산화 과정에서 k-means 클러스터링을 적용하여 그룹화한 후, 주어진 데이터가 원-핫 인코딩(one-hot encoding) 기법으로 원-핫 벡터(one-hot vector)로 변환한다. 이후, 특징 선택 기법 중 RFECV 기법을 활용하여 예측에 도움이 되는 변수를 가려내고, 이에 대해서만 변분 오토인코더를 활용하여 새로운 학습데이터를 생성한다. 제안 기법의 성능을 검증하기 위해 4가지 유형의 실험 데이터를 활용하여 데이터 증강 비율별로 그 유효성을 입증한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.2029-2032
• /
• 2004

The quantity of data were rapidly increased recently and caused the data overwhelming. This led to be difficult in searching the required data. The method of eliminating redundant data was needed. One of the efficient methods was Knowledge Discovery in Database (KDD). Generally data can be separate into 2 cases, continuous data and discrete data. This paper describes algorithm that transforms continuous attributes into discrete ones. We present an Improved Class Attribute Interdependence Maximization (ICAIM), which designed to work with supervised data, for discretized process. The algorithm does not require user to predefine the number of intervals. ICAIM improved CAIM by using significant test to determine which interval should be merged to one interval. Our goal is to generate a minimal number of discrete intervals and improve accuracy for classified class. We used iris plant dataset (IRIS) to test this algorithm compare with CAIM algorithm.

• 한국분말재료학회지
• /
• 제22권5호
• /
• pp.337-343
• /
• 2015

Powder compaction is a continually and rapidly evolving technology where it is a highly developed method of manufacturing reliable components. To understand existing mechanisms for compaction, parameter investigation is required. Experimental investigations on powder compaction process, followed by numerical modeling of compaction are presented in this paper. The experimental work explores compression characteristics of soft and hard ductile powder materials. In order to account for deformation, fracture and movement of the particles, a discrete-finite element analysis model is defined to reflect the experimental data and to enable investigations on mechanisms present at the particle level. Effects of important simulation factors and process parameters, such as particle count, time step, particle discretization, and particle size on the powder compaction procedure have been explored.

• 한국자동차공학회논문집
• /
• 제2권4호
• /
• pp.50-58
• /
• 1994

To examine the residual stress field resulting from stamping process for the lower control arm of a car, the explicit finite element analysis is performed for the stamping process by way of the ABAQUS Explicit. The residual stress is obtained in terms of the Von Mises stress and other parameters such as equivalent plastic strain, the change of blank thickness, the final configuration of the blank and the spring back effect are also considered. Moreover, discussed is the convergence of the explicit FEM versus the punch sped and the element discretization

• Coupled systems mechanics
• /
• 제4권3호
• /
• pp.211-235
• /
• 2015

In this work we present a one-dimensional damage model capable of representing the dynamic fracture for elastodamage bar with combined hardening in fracture process zone - FPZ and softening with embedded strong discontinuities. This model is compared with another one we recently introduced (Do et al. 2015) and it shows a good agreement between two models. Namely, it is indicated that strain-softening leads to a sensitivity of results on the mesh discretization. Strain tends to localization in a single element which is the smallest possible area in the finite element simulations. The strain-softening element in the middle of the bar undergoes intense deformation. Strain increases with increasing mesh refinement. Strain in elements outside the strain-softening element gradually decreases to zero.

• 대한전기학회논문지
• /
• 제37권5호
• /
• pp.282-288
• /
• 1988

A Multi-grid method is introduced to Finite Element Analysis of electromagnetic field problems in order to reduce the computational time. The puropse of this work is to study how to intermix discretization and solving process, thereby making the process more effective and to find the optimal factors of Multi-grid method. Several numerical experiments with linear models of uniform and nonuniform grids confirm that the proposed algorithm can reduce the computational time very effectively as compared with con ventional iterative methods. The best results are obtained with V cycle and S.O.R. with the acce leration factor of 1.3-1.4 for smoothing.