• Journal of KIISE:Databases
• /
• v.33 no.4
• /
• pp.372-383
• /
• 2006

Among the approaches to image database management, content-based image retrieval (CBIR) is viewed as having the best support for effective searching and browsing of large digital image libraries. Typical CBIR systems allow a user to provide a query image, from which low-level features are extracted and used to find 'similar' images in a database. However, there exists the semantic gap between human visual perception and low-level representations. An effective methodology for overcoming this semantic gap involves relevance feedback to perform feature re-weighting. Current approaches to feature re-weighting require the number of components for a feature representation to be the same for every image in consideration. Following this assumption, they map each component to an axis in the n-dimensional space, which we call the metric space; likewise the feature representation is stored in a fixed-length vector. However, with the emergence of features that do not have a fixed number of components in their representation, existing feature re-weighting approaches are invalidated. In this paper we propose a feature re-weighting technique that supports features regardless of whether or not they can be mapped into a metric space. Our approach analyses the feature distances calculated between the query image and the images in the database. Two-sided confidence intervals are used with the distances to obtain the information for feature re-weighting. There is no restriction on how the distances are calculated for each feature. This provides freedom for how feature representations are structured, i.e. there is no requirement for features to be represented in fixed-length vectors or metric space. Our experimental results show the effectiveness of our approach and in a comparison with other work, we can see how it outperforms previous work.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2000.04a
• /
• pp.398-413
• /
• 2000

본 논문의 목적은 유사도와 강화 학습을 사용하여, 정보를 제공하는 에이전트와 정보를 요청하는 에이전트간의 연결을 매개하는 조정 에이전트(Coordination Agent, Middle Agent) 구현 방식을 제안하는데 있다.본 논문에서는 질의 에이전트의 질의와 가장 밀접한 정보를 제공하는 것으로 판단되는 정보 에이전트를 찾는 방안을 제안하고자 한다. 정보 에이저트와 질의 에이전트는 조정에이전트에 정보를 등록·요청할 때, 조정 에이전트에 이미 존재하는 기본 오톨로지(Base Ontology)에 자신이 제공·질의하는 정보의 상대적 가중치를 함께 등록한다. 조정 에이전트는 질의 에이전트와 정보 에이전트의 가중치를 고려하여 유사도를 구하고, 구해진 유사도를 이용하여 가장 근접한 정보를 제공하는 정보 에이전트를 찾아 연결한다. 가중치를 제공하지 않는 질의 에이전트의 경우에는 강화 학습으로 얻어진 특성 자료를 이용하여 조정 에이전튼가 임의로 가중치를 구하고, 얻어진 결과에 대하여 타당성을 검증한다.

• The Korean Journal of Applied Statistics
• /
• v.34 no.3
• /
• pp.427-438
• /
• 2021

Two-stage sampling allows us to estimate population characteristics by both unit and cluster level together. Given a complex auxiliary information, integrated calibration weighting would better reflect the level-wise characteristics as well as multivariate characteristics between levels. This paper explored the integrated calibration weighting methods by Estevao and Särndal (2006) and Kim (2019) through a simulation study, where the efficiency of those weighting methods was compared using an artificial population data. Two weighting methods among others are shown efficient: single step calibration at the unit level with stacked individualized auxiliary information and iterative integrated calibration at each level. Under both methods, cluster calibrated weights are defined as the average of the calibrated weights of the unit(s) within cluster. Both were very good in terms of the goodness-of-fit of estimating the population totals of mutual auxiliary information between clusters and units, and showed small relative bias and relative mean square root errors for estimating the population totals of survey variables that are not included in calibration adjustments.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.7
• /
• pp.9-15
• /
• 2009

The fuzzy weighted mean classifier is one of the most common classification models and could achieve high performance by adjusting the weights. However, the weights were generally decided based on the experience of experts, which made the resulting classifiers to suffer the lack of consistency and objectivity. To resolve this problem, in this paper, a weight deciding method based on the statistics of the data is introduced, which ensures the learned classifiers to be consistent and objective. To investigate the effectiveness of the proposed methods, Iris data set available from UCI machine learning repository is used and promising results are obtained.

• The Korean Journal of Applied Statistics
• /
• v.36 no.5
• /
• pp.381-397
• /
• 2023

The most representative design used in clinical trials is randomization, which is used to accurately estimate the treatment effect. However, comparison between the treatment group and the control group in an observational study without randomization is biased due to various unadjusted differences, such as characteristics between patients. Propensity score weighting is a widely used method to address these problems and to minimize bias by adjusting those confounding and assess treatment effects. Inverse probability weighting, the most popular method, assigns weights that are proportional to the inverse of the conditional probability of receiving a specific treatment assignment, given observed covariates. However, this method is often suffered by extreme propensity scores, resulting in biased estimates and excessive variance. Several alternative methods including trimming, overlap weights, and matching weights have been proposed to mitigate these issues. In this paper, we conduct a simulation study to compare performance of various propensity score weighting methods under diverse situation, such as limited overlap, misspecified propensity score, and treatment contrary to prediction. From the simulation results overlap weights and matching weights consistently outperform inverse probability weighting and trimming in terms of bias, root mean squared error and coverage probability.

• Journal of Navigation and Port Research
• /
• v.29 no.1 s.97
• /
• pp.59-64
• /
• 2005

MDMG(Multiple Decision-Making Group) problems comprise those of UDMG(Unit Decision-Making Group) which contradict each other. For the evaluation problem of port competitiveness, it has the complicated evaluation characteristics of multi-strata-complex and multi-attributes. Especially, it becomes typical MDMG problems in the evaluation which a great number of decision makers such as shipping companies, freight forwarders, logistics companies and researchers participate in This evaluation of complex problems needs the compensated process of weight which rationally unites heterogeneous preferences of each of groups. In this respect, the purpose of this study is to remove the uncertainty of the UDMG using the theory of DS (Dempster-Shafer) and present the integrated weight through the level process.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.5 no.3
• /
• pp.533-541
• /
• 2001

In this paper, we proposed lowpass filter using all-pass sums of flat delay characteristics. this filter consisted of all-pass filter of parallel structure, the general analog filter is impossible to adjust the phase and the delay, using the Proposed filter, it has advantage to adjust them. And, we compared and analyzed this filter with passband width and magnitude characteristics, and the relation of group delay characteristics and cut-off frequency. Also, in order to obtain desired cut-off frequency, forming the weighing, we obtained desired cut-off frequency and group delay characteristics.

• Journal of Broadcast Engineering
• /
• v.22 no.3
• /
• pp.327-338
• /
• 2017

Exposure fusion is a typical image fusion technique to generate a high dynamic range image by combining two or more different exposure images. In this paper, we propose block-based exposure adjustment considering unique characteristics of human visual system and improved saturation measure to get weight map. Proposed exposure adjustment artificially corrects intensity values of each input images considering human visual system, efficiently preserving details in the result image of exposure fusion. The improved saturation measure is used to make a weight map that effectively reflects the saturation region in the input images. We show the superiority of the proposed algorithm through subjective image quality, MEF-SSIM, and execution time comparison with the conventional exposure fusion algorithm.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.05a
• /
• pp.307-310
• /
• 2003

본 논문에서는 대표적인 비선형 동특성을 가지는 실제 헬리콥터의 회전 띤 자세 운동을 근사화한 모형 헬리콥터의 시스템을 소개하고 이 시스템의 정지 자세 제어를 목표로 직접 적응 웨이브렛 신경회로망 제어기를 다음의 과정에 의해 만든다. 우선 상태 공간에 적용할 웨이브렛 기준 함수를 정의하고 나서 제어기로 들어오는 입력 값의 대략적인 범위와 특성을 파악해서 웨이브렛 이론에 근거해 신축(dilation)과 이동(traslation) 변수 값을 선택하여 초기 적응 웨이브렛 신경회로망 제어기를 건설한다. 마지막으로 시스템의 안정화 제어를 위하여 선택, 교배, 돌연변이의 진화연산자에 의해 일시에 최적의 구조와 결합가중치로 진화시켜 가는 새로운 형태의 ENNC를 제안하여 연결 가중치(weight)를 조정한다. 이 직접 적응 웨이브렛 신경회로망 제어기를 비선형 시스템인 모형 헬리콥터 시뮬레이터에 적용하여 제안한 제어기의 견실성 및 그 우수성을 입증하고자 한다.

• 한국HCI학회:학술대회논문집
• /
• 2008.02a
• /
• pp.445-450
• /
• 2008

본 논문에서는 객체의 인식성을 높이기 위해 가상의 지역 광원을 사용한 카툰 렌더링 기법을 제안한다. 지역 광원은 각 메쉬의 기하 정보를 분석하여 배치되며 객체의 지역적인 음영 대비를 증가시켜 객체의 모양과 특정이 눈에 잘 띄도록 한다. 하지만 지역 광원을 사용한 쉐이딩 기법은 객체의 일부 영역에서 갈라지고 불연속적인 음영을 만들어 이미지의 질적 하락을 초래한다. 이러한 현상을 막기 위하여 곡률과 샐리언시의 개념을 사용하여 영역의 특성에 따라 차등적으로 지역 광원을 객체에 적용하였다. 곡률은 해당 영역의 기하적 특성을 구분하여 지역 광원에 의한 음영 대비 증감을 조정하고, 샐리언시는 영역의 중요도를 판별하여 곡률이 쉐이딩에 미치는 가중치를 조절한다.