• ETRI Journal
• /
• 제34권4호
• /
• pp.564-571
• /
• 2012

This paper presents an algorithm for removing high-density impulsive noise that generates some serious distortions in edge regions of an image. Although many works have been presented to reduce edge distortions, these existing methods cannot sufficiently restore distorted edges in images with large amounts of impulsive noise. To solve this problem, this paper proposes a method using connected lines extracted from a binarized image, which segments an image into uniform and edge regions. For uniform regions, the existing simple adaptive median filter is applied to remove impulsive noise, and, for edge regions, a prediction filter and a line-weighted median filter using the connected lines are proposed. Simulation results show that the proposed method provides much better performance in restoring distorted edges than existing methods provide. When noise content is more than 20 percent, existing algorithms result in severe edge distortions, while the proposed algorithm can reconstruct edge regions similar to those of the original image.

• 전자공학회논문지B
• /
• 제32B권5호
• /
• pp.812-820
• /
• 1995

This paper proposes an approach of identifying nonlinear mappings from input/output data. The approach is based on the universal approximation by the fuzzy integration of local affine mappings. A connectionist model realizing the universal approximator is suggested by using a processing unit based on both the radial basis function and the weighted sum scheme. In addition, a learning method with self-organizing capability is proposed for the identifying of nonlinear mapping relationships with the given input/output data. To show the effectiveness of our approach, the proposed model is applied to the function approximation and the prediction of Mackey-Glass chaotic time series, and the performances are compared with other approaches.

• Communications for Statistical Applications and Methods
• /
• 제16권4호
• /
• pp.647-658
• /
• 2009

It is already known from the previous study that flood seems to have heavier tail. Therefore, to make prediction of future extreme label, some agreement of tail behavior of extreme data is highly required. The LH-moments estimation method, the generalized form of L-moments is an useful method of characterizing the upper part of the distribution. LH-moments are based on linear combination of higher order statistics. In this study, we have formulated LH-moments of five distributions useful in hydrology such as, two types of three parameter kappa distributions, beta-${\kappa}$ distribution, beta-p distribution and a generalized Gumbel distribution. Using LH-moments reduces the undue influences that small sample may have on the estimation of large return period events.

• 품질경영학회지
• /
• 제33권1호
• /
• pp.73-82
• /
• 2005

This article proposes a practical approach, which is based on the concept of the expected squared relative error, that can consider both the prediction quality and the practitioner's subjectivity in simultaneously optimizing multiple responses. Through a case study, multiresponse optimization using the expected squared relative error approach is illustrated, and the SAS program to implement the proposed method is provided.

• 국제초고층학회논문집
• /
• 제7권2호
• /
• pp.171-183
• /
• 2018

Turner and Penn (1) from UCL have proved that Visibility Graph Analysis (VGA) can be used as a more accurate method to predict the pedestrian distribution in building spaces. However, this methodology neglects certain elements that are of special influence on pedestrian distribution in buildings, especially the entrances and exits. Based on Space Syntax, this dissertation improves on the traditional method of Visibility Graph Analysis, using three shopping centers in Shenzhen as examples, attempts to explore a new parameter - "attenuation index of pedestrians at the entrances and exits" - using relevant data of the entrances and exits of the three cases, and combines it with traditional VGA analysis through weighted calculation, in order to provide more accurate predictions of pedestrian patterns in shopping centers.

• Computers and Concrete
• /
• 제11권6호
• /
• pp.603-620
• /
• 2013

This paper improves genetic programming (GP) and weight genetic programming (WGP) and proposes soft-computing polynomials (SCP) for accurate prediction and visible polynomials. The proposed genetic programming system (GPS) comprises GP, WGP and SCP. To represent confined compressive strength and strain of circular concrete columns in meaningful representations, this paper conducts sensitivity analysis and applies pruning techniques. Analytical results demonstrate that all proposed models perform well in achieving good accuracy and visible formulas; notably, SCP can model problems in polynomial forms. Finally, concrete compressive strength and lateral steel ratio are identified as important to both confined compressive strength and strain of circular concrete columns. By using the suggested formulas, calculations are more accurate than those of analytical models. Moreover, a formula is applied for confined compressive strength based on current data and achieves accuracy comparable to that of neural networks.

• 소음진동
• /
• 제10권1호
• /
• pp.57-63
• /
• 2000

On the basis of the concept of strain energy-weighted dissipation, an enhanced model for predicting damping in laminates is presented. In this model, the influence of transverse shear on $90^{\circ}$ laminates has been included with those of in-plane stresses on beam. Also, an experimental damping measurement is conducted with changing the length and the thickness of laminated beam specimen for confirmation of the model prediction. The theoretical predictions in $90^{\circ}$laminates were reasonably compared with experimental data. The transverse shear reveals to have an influence on the damping, behavior in $90^{\circ}$ laminates.

• Computers and Concrete
• /
• 제23권4호
• /
• pp.255-265
• /
• 2019

The potential of using genetic programming to predict engineering data has caught the attention of researchers in recent years. The present paper utilized weighted genetic programming (WGP), a derivative model of genetic programming (GP), to model the compressive strength of concrete. The calculation results of Abrams' laws, which are used as the design codes for calculating the compressive strength of concrete, were treated as the inputs for the genetic programming model. Therefore, knowledge of the Abrams' laws, which is not a factor of influence on common data-based learning approaches, was considered to be a potential factor affecting genetic programming models. Significant outcomes of this work include: 1) the employed design codes positively affected the prediction accuracy of modeling the compressive strength of concrete; 2) a new equation was suggested to replace the design code for predicting concrete strength; and 3) common data-based learning approaches were evolved into knowledge-based learning approaches using historical data and design codes.

• International journal of advanced smart convergence
• /
• 제12권2호
• /
• pp.17-24
• /
• 2023

Given the surge in the elderly population, and increasing in dementia cases, there is a growing interest in digital therapies that facilitate steady remote treatment. However, in the cognitive assessment of digital therapies through clinical trials, the absence of log data as an essential evaluation factor is a significant issue. To address this, we propose a solution of utilizing weighted derived variables based on high-importance variables' accuracy in log data utilization as an indirect cognitive assessment factor for digital therapies. We have validated the effectiveness of this approach using machine learning techniques such as XGBoost, LGBM, and CatBoost. Thus, we suggest the use of log data as a rapid and indirect cognitive evaluation factor for digital therapy users.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2024년도 춘계학술발표대회
• /
• pp.784-785
• /
• 2024

형량 예측 연구는 법률 인공지능에서 가장 활발히 연구되고 있는 분야 중 하나이며, 비법률전문가의 사법 신뢰도 상승과 법률전문가의 업무 부담 완화에 긍정적 영향을 줄 수 있다. 본 연구는 형사 사건의 양형 예측에 ChatGPT 를 접목하여 입력된 사실관계와 유사한 선행 판례를 검색함으로써 형량 예측에 필요한 모델의 훈련 시간과 비용을 절감하는 접근법을 제안한다. 본 모델의 weighted F1-score 는 0.53 으로, 미세조정된 BERT 모델과 유사한 성능을 기록하였다.