Combining Linear and Nonlinear Normalization Methods to Improve Learning Performance in Traffic Scene Recognition

김대현 사단법인 한국융합기술연구학회 2023 아시아태평양융합연구교류논문지 Vol.9 No.1

Although input vector properties are the main determinants of neural network performance, most studies use a simple linear scaling model to normalize the input vector without considering input vector normalization. Because single normalization maps only a single property of the input data, the current normalization approach may not be effective for improving prediction accuracy. In addition, a linear type of normalization is just a linear mapping that mirrors raw data properties. On the other hand, non-linear normalization can reflect non-linearly by emphasizing certain properties. This study aims to propose a new and efficient normalization method that can provide better prediction performance by learning machines. For this purpose, various regularization methods such as linear and nonlinear mapping models and a combination of the two models were considered. The research methodology is as follows. a) Based on theoretical studies on linear and nonlinear models of regularization methods, a method that combines two different normalization methods is proposed. b) Data are prepared for experiments in neural network models. c) Experiments are conducted with the neural network model on two network topologies with three different normalization methods and 30 trials with different initial weights, and the results are obtained. d) The prediction performance is compared, and the best normalization method for traffic scene analysis is suggested. Experimental studies have shown that the prediction performance of the input vector by the nonlinear normalization method is 0.6–2.4% better than the prediction performance of the input vector by the linear method. More importantly, the predictive performance of the combination of the two normalization methods is 2.4–7.2% better than the linear method and 2.8–4.0% better than the non-linear method. The results of this study revealed that the combined use of two different normalization methods provides better performance in machine learning compared to individual normalization methods, such as linear and nonlinear normalization on input vectors.

심층신경망 기반의 음성인식을 위한 절충된 특징 정규화 방식

김민식(Min Sik Kim),김형순(Hyung Soon Kim) 한국음성학회 2020 말소리와 음성과학 Vol.12 No.3

특징 정규화는 음성 특징 파라미터들의 통계적인 특성의 정규화를 통해 훈련 및 테스트 조건 사이의 환경 불일치의 영향을 감소시키는 방법으로서 기존의 Gaussian mixture model-hidden Markov model(GMM-HMM) 기반의 음성인식 시스템에서 우수한 성능개선을 입증한 바 있다. 하지만 심층신경망(deep neural network, DNN) 기반의 음성인식 시스템에서는 환경 불일치의 영향을 최소화 하는 것이 반드시 최고의 성능 개선으로 연결되지는 않는다. 본 논문에서는 이러한 현상의 원인을 과도한 특징 정규화로 인한 정보손실 때문이라 보고, 음향모델을 훈련 하는데 유용한 정보는 보존하면서 환경 불일치의 영향은 적절히 감소시켜 음성인식 성능을 최대화 하는 특징 정규화 방식이 있는지 검토해보고자 한다. 이를 위해 평균 정규화(mean normalization, MN)와 평균 및 분산 정규화(mean and variance normalization, MVN)의 절충 방식인 평균 및 지수적 분산 정규화(mean and exponentiated variance normalization, MEVN)를 도입하여, 잡음 및 잔향 환경에서 분산에 대한 정규화의 정도에 따른 DNN 기반의 음성인식 시스템의 성능을 비교한다. 실험 결과, 성능 개선의 폭이 크지는 않으나 분산 정규화의 정도에 따라 MEVN이 MN과 MVN보다 성능이 우수함을 보여준다. Feature normalization is a method to reduce the effect of environmental mismatch between the training and test conditions through the normalization of statistical characteristics of acoustic feature parameters. It demonstrates excellent performance improvement in the traditional Gaussian mixture model-hidden Markov model (GMM-HMM)-based speech recognition system. However, in a deep neural network (DNN)-based speech recognition system, minimizing the effects of environmental mismatch does not necessarily lead to the best performance improvement. In this paper, we attribute the cause of this phenomenon to information loss due to excessive feature normalization. We investigate whether there is a feature normalization method that maximizes the speech recognition performance by properly reducing the impact of environmental mismatch, while preserving useful information for training acoustic models. To this end, we introduce the mean and exponentiated variance normalization (MEVN), which is a compromise between the mean normalization (MN) and the mean and variance normalization (MVN), and compare the performance of DNN-based speech recognition system in noisy and reverberant environments according to the degree of variance normalization. Experimental results reveal that a slight performance improvement is obtained with the MEVN over the MN and the MVN, depending on the degree of variance normalization.

A Computed Tomography-Based Spatial Normalization for the Analysis of [18F] Fluorodeoxyglucose Positron Emission Tomography of the Brain

조한나,Jin Su Kim,Jae Yong Choi,유영훈,류철형 대한영상의학회 2014 Korean Journal of Radiology Vol.15 No.6

Objective: We developed a new computed tomography (CT)-based spatial normalization method and CT template to demonstrate its usefulness in spatial normalization of positron emission tomography (PET) images with [18F] fluorodeoxyglucose (FDG) PET studies in healthy controls. Materials and Methods: Seventy healthy controls underwent brain CT scan (120 KeV, 180 mAs, and 3 mm of thickness) and [18F] FDG PET scans using a PET/CT scanner. T1-weighted magnetic resonance (MR) images were acquired for all subjects. By averaging skull-stripped and spatially-normalized MR and CT images, we created skull-stripped MR and CT templates for spatial normalization. The skull-stripped MR and CT images were spatially normalized to each structural template. PET images were spatially normalized by applying spatial transformation parameters to normalize skull-stripped MR and CT images. A conventional perfusion PET template was used for PET-based spatial normalization. Regional standardized uptake values (SUV) measured by overlaying the template volume of interest (VOI) were compared to those measured with FreeSurfer-generated VOI (FSVOI). Results: All three spatial normalization methods underestimated regional SUV values by 0.3–20% compared to those measured with FSVOI. The CT-based method showed slightly greater underestimation bias. Regional SUV values derived from all three spatial normalization methods were correlated significantly (p < 0.0001) with those measured with FSVOI. Conclusion: CT-based spatial normalization may be an alternative method for structure-based spatial normalization of [18F] FDG PET when MR imaging is unavailable. Therefore, it is useful for PET/CT studies with various radiotracers whose uptake is expected to be limited to specific brain regions or highly variable within study population.

사학분쟁조정위원회의 학교법인 정상화에 관한 법리

김명연(Kim, Myoung-Youn) 한국비교공법학회 2017 공법학연구 Vol.18 No.2

사학분쟁조정위원회는 사회민주화 과정에서 퇴출되었던 사학비리운영자를 예외 없이 복귀시켰다. 사학분쟁조정위원회가 사회의 논란 속에 비리재단을 복귀시킨 상지대 정상화 결정이 대법원 이사선임처분 취소판결에 의하여 취소되었다. 교육부는 이 판결이 다른 사립학교에 미치는 영향을 우려하여 이를 개방이사 선임과 관련한 절차상 하자로 의미를 축소하고 있다. 그리고 종래의 정상화 심의원칙에 따라 상지대를 신속하게 재정상화하려고 하고 있다. 그러나 이 판결로 인한 재정상화 심의에서는 개방이사 선임주체, 개방이사제도와 관련한 임시이사의 권한의 한계, 개방이사 선임을 예정하지 않은 정상화 심의원칙의 변경 등 다양한 법적 쟁점이 있다. 또한 상지대 정상화 결정 이후 헌법재판소와 대법원에 의하여 종래 정상화 심의원칙의 핵심적 법리를 모두 부정하는 새로운 정상화법리가 형성되었다. 이 논문에서는 상지대 재정상화 심의와 관련하여 이사선임처분 취소판결의 의의와 내용, 정상화의 요건과 재정상화 심의의결의 적법성, 새로운 정상화법리의 형성과 정상화 심의원칙의 변경, 개방이사 선임과 관련한 법적 문제, 종전이사의 법적 지위와 이사추천권 부여의 적격성 심사 등의 법적 쟁점을 분석하고 이에 관한 법리를 제시하고 있다. 2017년 4월 사학분쟁조정위원회가 정상화한 62개 사립학교 중 56개 학교에 상지대와 같이 개방이사를 선임하지 않은 위법한 정상화 결정이 이루어졌다. 상지대 재정상화 관련한 법적 쟁점과 법리는 이들 학교의 재정상화와 정상화 심의가 예정된 적지 않은 사립학교에도 타당하게 적용될 수 있다. 사학비리의 근절은 교육의 본질과 관련된 근본문제이며, 교육의 정상화를 위한 선결과제이다. 사학분쟁조정위원회가 비리사학운영자를 학교운영에서 배제하여 사학의 비정상화의 정상화에 대한 법적·사회적 책무를 다할 것을 기대한다. Private School Dispute Resolution Committee(PSDR) brings back private school interested groups without any exception, who were exiled through the process on social democracy movement into schools. Supreme Court, eventually, revoked this normalization of school made by PSDR, which was revocation judgement of disposition of directors’ appointment. The Ministry of Education intentionally distorts the Supreme Court’s decision by arguing that the decision just exposes procedural errors on appointing public director. It is because the Ministry of Education concerns about uncertain impacts on the other private schools by the decision. As the result, the Ministry of Education is trying to renormalize Sangji University based on the precedent deliberation principles of normalization. However, the Supreme Court’s decision exposes plenty of legal issues such as who appoints public directors, temporary director’s authority and limit about public directors, and deliberation principles of normalization which does not notify employment of public directors in advance on the way of the normalization. This paper analyzes legal issues and suggests legal principles associated with these such as the meaning and significance of revocation judgement of directors’ appointment about Sangji University’s normalization, requirements for the normalization and legality of deliberation principles of normalization, development of modified legal principles for normalization, legal issue on employing public director, and then previous director’s legal status and its eligibility for right of recommending other directors. At April, 2017, illegal normalization decision, which did not appointment of public director as Sangji University did, were passed off to 56 private schools among 62 schools that PSDR normalized. Therefore, the legal issues and principles associated with Sangji University’s renormalization would be lawfully applied to other private schools appointed deliberation of normalization or renormalization. Extermination of private schools’ corruption is fundamental and essential issues about education. Furthermore, it must be done as the first priority for normalizing education. Private School Dispute Resolution Committee should fulfill its legal and social responsibility to eradicate the private schools’ corruption.

An Iterative Normalization Algorithm for cDNA Microarray Medical Data Analysis

Kim, Yoonhee,Park, Woong-Yang,Kim, Ho Korea Genome Organization 2004 Genomics & informatics Vol.2 No.2

A cDNA microarray experiment is one of the most useful high-throughput experiments in medical informatics for monitoring gene expression levels. Statistical analysis with a cDNA microarray medical data requires a normalization procedure to reduce the systematic errors that are impossible to control by the experimental conditions. Despite the variety of normalization methods, this. paper suggests a more general and synthetic normalization algorithm with a control gene set based on previous studies of normalization. Iterative normalization method was used to select and include a new control gene set among the whole genes iteratively at every step of the normalization calculation initiated with the housekeeping genes. The objective of this iterative normalization was to maintain the pattern of the original data and to keep the gene expression levels stable. Spatial plots, M&A (ratio and average values of the intensity) plots and box plots showed a convergence to zero of the mean across all genes graphically after applying our iterative normalization. The practicability of the algorithm was demonstrated by applying our method to the data for the human photo aging study.

Korea-Japan Relations during the Second Republic of Korea: A Counterfactual Analysis of the Aborted Korea-Japan Normalization

윤태룡 사단법인 한국평화연구학회 2009 평화학연구 Vol.10 No.2

This paper purports to provide a counterfactual analysis of a non-event during the Second Republic of Korea, that is, the aborted Korea-Japan Normalization. The author argues that without General Park Chung Hee’s military coup on May 16, 1961, the rapprochement between the two former enemies would have been possible during the Second Republic of Korea. The ongoing process of negotiations for diplomatic normalization between South Korea and Japan would have been accelerated and have reached the ultimate goal but for the military’s unconstitutional toppling down of the first democratic regime in South Korea. Many experts argue that President Park’s strong leadership or the U.S. pressure is the main reason for the achievement of Korea-Japan Normalization. However, based on counterfactual analysis, this paper concludes that Korea-Japan Normalization would have been achieved much earlier than the year of 1965 when the two states actually reached an agreement only after spending four years of tough bargaining after the coup in South Korea. This paper purports to provide a counterfactual analysis of a non-event during the Second Republic of Korea, that is, the aborted Korea-Japan Normalization. The author argues that without General Park Chung Hee’s military coup on May 16, 1961, the rapprochement between the two former enemies would have been possible during the Second Republic of Korea. The ongoing process of negotiations for diplomatic normalization between South Korea and Japan would have been accelerated and have reached the ultimate goal but for the military’s unconstitutional toppling down of the first democratic regime in South Korea. Many experts argue that President Park’s strong leadership or the U.S. pressure is the main reason for the achievement of Korea-Japan Normalization. However, based on counterfactual analysis, this paper concludes that Korea-Japan Normalization would have been achieved much earlier than the year of 1965 when the two states actually reached an agreement only after spending four years of tough bargaining after the coup in South Korea.

임시이사 선임 학교법인의 정상화 사례 분석과 시사점

주영달 ( Ju Young Dal ) 대한교육법학회 2018 敎育 法學 硏究 Vol.30 No.1

공교육에서 비중이 큰 사학에서 부정과 분쟁이 발생하여 학교법인과 사립학교가 정상적으로 운영되지 못하는 경우에는 국가가 관여하여 임시이사를 선임하여 정상화를 추진하게 된다. 그런데 정이사 선임이라는 정상화가 이루어지기도 어렵고 정상화가 이루어진 이후에도 지속적으로 사학에 파행이 발생하는 경우가 많이 있다. 이처럼 신속하고 효과적인 정상화가 이루어지지 않는 원인을 파악함으로써 정상화 제도 개선에 대한 시사점을 얻기 위하여 대학을 중심으로 사학의 분쟁이나 부정이 발생하여 정상화 과정을 거쳤던 학교법인의 사례를 분석하여 보았다. 사학분쟁조정위원회의 자료집을 기초로 분쟁이 심각하거나 장기간 지속되었던 학교법인, 분쟁이 재발하였던 학교법인, 원만한 정상화를 이루었던 학교법인을 선정한 다음 각 대학에 대한 언론 보도를 분석하였다. 임시 이사가 선임된 학교법인의 사례를 분석한 결과 정상화 과정에서의 임원취임승인취소제도, 정상화 시기 판단과 그 주체, 정상화 방법 또는 정상화 제도, 누구의 의사를 반영하여 누구를 정이사로 선임할 것인가의 정상화 내용 등에 관한 문제가 정상화에서 주된 연구과제임이 확인되었다. Today's private schools are incorporated into the public education system and function as public education institutions, and the proportion of public education is very high. Therefore, if the school foundation and the private school can not operate properly due to the negligence and disputes in the history of the school, the state is involved and appoints temporary directors to promote normalization. However, normalization of the appointment is difficult to achieve, and even after the normalization process, there are many cases in which there is a abnormality in history. Therefore, we tried to analyze the case of the school foundation which had undergone normalization process due to the dispute or negation of the private school centering on the university. As a result of analyzing the case of the school corporation in which the temporary director has been appointed, it is found that the problem of the cancellation of the approval of the inauguration of officers in normalization process, the subject of the judgment of normalization period, normalization system, normalization of who will be elected as a righteous person by reflecting their opinions. And these are the main research task in normalization.

Variant Call Format Normalization Contributes to the Accuracy of Variant Nomenclature

Eunkyoung You,Jong-Ho Park,박창훈,박경진 대한임상검사정도관리협회 2022 Journal of Laboratory Medicine And Quality Assuran Vol.44 No.4

Background: The variant call format (VCF) is currently used to store information on genetic variants. Variant description from the VCF is not unique or standardized. A recent survey indicated that more than half of the clinical laboratories in Korea have not performed VCF normalization procedures in their bioinformatics pipeline during routine analysis. We evaluated the effect of variant normalization on variant nomenclature using two tools. Methods: Two binary alignment map files (GNG-21-01 and GNG-21-04) and validated variant descriptions (single-nucleotide variations [n=51] and insertions/duplications/deletions [n=13]) were provided by the Korean Association of External Quality Assessment Service. Variant calling and annotation were performed using Torrent Suite and snpEff, respectively. The vt-normalize and Genome Analysis Toolkit (GATK) LeftAlignAndTrimVariants (LAATV) were used to evaluate the effect of VCF normalization. Results: The accuracy of the nomenclature of un-normalized variants was 99% (63/64); one variant of NM_173630.3(RTTN ): c.725_727dupGAG was incorrectly described as c.724_726dupGGA. The variant description was corrected through VCF normalization. The normalization results obtained using vt-normalize and GATK LAATV were comparable. Conclusions: VCF normalization may not generate completely correct nomenclature but contributes to more accurate annotation of variants. VCF normalization should be introduced into bioinformatics pipelines in clinical laboratories to ensure more reliable annotations of variants.

푸코 윤리학의 관점에서 조명한정상성의 본질 연구

김분선 한국하이데거학회 2024 현대유럽철학연구 Vol.- No.72

이 글은 푸코의 윤리학에서 등장하는 철학적 치유의 개념이 캉킬렘의 연구에 빚지고 있음을 해명할 것이다. 나는 이 글에서 캉킬렘이 생물학과 병리학을 구별하는 주된 기준으로 ‘정상’의 개념들을 활용하고 있다는 점에 주목하여 정상(normal), 정상성(normality), 정상화(normalization)의 기초적인 개념들을 파악한다. 이를 바탕으로 정상과 비정상에 관한 개념들이 가치적 양상을 띠고 있고 규범적인 문제들과 관련이 있으며 관계적인 측면에서 고정되는 질서라는 특이점이 있음을 조명한다. 더 나아가 현대 병리학의 비정상성의 문제에 대한 해석, 칸트의 인간학에 등장하는 정상성의 문제들을 조망한 후, 푸코의 자기 배려의 윤리에서 다루고 있는 철학의 자기 치유에 관한 논의들을 분석한다. 결과적으로 나는 푸코의 윤리학에서 쓰이는 체현하는 윤리 주체의 배려활동이 자기 정상화의 과정이며 정상성을 회복하려는 윤리적 기술 함양의 과정에 속함을 주장할 것이다. This paper argues that the philosophical concept of cure in Foucault’s ethics is based on Canquilem’s research. I note that Canquilem used the construct of normal as the main criterion for distinguishing biology from pathology. In this regard, I explore the basic concepts of normal, normality, and normalization before clarifying that the notions of normal and abnormal have a value aspect and are related to normative issues. I indicate that normal and abnormal have a unique point as a sequence in the relational aspect. On the basis of the preceding discussions, I analyze the interpretation of the problem of abnormality in modern pathology and the problems of normality that appear in Kant’s anthropology. I subsequently review discussions of self-healing in philosophy that are addressed in Foucault’s ethics of self-care. These culminate in my assertion that the possibility of an ethical subject of embodiment, as used in Foucault’s ethics, arises from the process of caring activities as a process of self-normalization and the cultivation of ethical skills to restore normality.

음성인식에서 화자 내 정규화를 위한 진폭 변경 방법

김동현 ( Dong-hyun Kim ),홍광석 ( Kwang-seok Hong ) 한국인터넷정보학회 2003 인터넷정보학회논문지 Vol.4 No.3

기존의 성도 정규화 방법은 화자 간 정규화의 정확성을 개선하기 위한 매우 좋은 방법이다. 본 논문에서는 피치 변경 발성에 기반을 둔 새로운 화자 내 warping 인수 추정 방법을 제안한다. 화자 내 피치 변경 발성은 성문과 성도에 의해 발생되는 음성의 음향학적 차이 때문에 음성의 특징 공간 분포는 다르게 나타날 것이다. 발성의 변동은 frequency 성분과 amplitude 성분의 두가지 유형이 있다. 성도 정규화는 화자 간 정규화 방법들 중에서 주파수 정규화 방법이다. 여기에서는 화자 내 정규화를 위하여 진폭 변동을 정규화하는 방법을 제안한다. 참조 피치와 입력 피치의 역비례 계산에 의해서 진폭 warping 인수를 결정하는 것이 가능하다. 성능 평가를 위한 인식 실험 결과 숫자와 단어 인식에서 0.4%∼2.3% 정도의 인식 오류가 감소되었다. The method of vocal tract normalization is a successful method for improving the accuracy of inter-speaker normalization. In this paper, we present an intra-speaker warping factor estimation based on pitch alteration utterance. The feature space distributions of untransformed speech from the pitch alteration utterance of intra-speaker would vary due to the acoustic differences of speech produced by glottis and vocal tract. The variation of utterance is two types: frequency and amplitude variation. The vocal tract normalization is frequency normalization among inter-speaker normalization methods. Therefore, we have to consider amplitude variation, and it may be possible to determine the amplitude warping factor by calculating the inverse ratio of input to reference pitch. As the recognition results, the error rate is reduced from 0.4% to 2.3% for digit and word decoding.