• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2003년도 ISIS 2003
• /
• pp.162-165
• /
• 2003

Differential learning relies on the differentiated values of nodes, whereas the conventional learning depends on the values themselves of nodes. In this paper, I elucidate the differential learning in the framework maximum likelihood learning of linear generative model with latent variables obeying random walk. I apply the idea of differential learning to the problem independent component analysis(ICA), which leads to differential ICA. Algorithm derivation using the natural gradient and local stability analysis are provided. Usefulness of the algorithm is emphasized in the case of blind separation of temporally correlated sources and is demonstrated through a simple numerical example.

• 대한의용생체공학회:의공학회지
• /
• 제29권1호
• /
• pp.73-81
• /
• 2008

The purpose of this study is to observe the brain activity patterns during visual oddball tasks with two difficulty levels by the analysis of high-density event-related potential (ERP). Along with conventional statistical analysis of averaged ERP waveforms, we applied independent component analysis (ICA) for the individual, single-trial analysis and verified its effectiveness. We could identify multiple ERP components such as early visual components (P1, N1), and two components which seem to be important task-related components and showed difficulty-dependent variability (P2, P300). The P2 was found around central region at $180{\sim}220ms$, and the P300 was found globally at $300{\sim}500ms$ poststimulus. As the task became difficult, the P2 amplitude increased, and the P300 amplitude decreased. After single-trial ERPs were decomposed into multiple independent components (ICs), several ICs resulting from P2 and P300 sources were identified. These ICs were projected onto scalp electrodes and the projected ICs were statistically compared according to two task difficulties. For most subjects, the results obtained from single-trial/individual analysis using ICA gave the tendencies of amplitude change that are similar to the averaged ERP analysis for most subjects. The temporal pattern and number of ICs corresponding to ${\mu}$ rhythm was not dependent on the task difficulty. It seems that the motor response was not affected by the task difficulty.

• 응용통계연구
• /
• 제30권2호
• /
• pp.271-280
• /
• 2017

우리나라 뉴스에서 매일 빠지지 않는 내용은 아마도 부동산 경제에 관한 것이라고 생각된다. 많은 사람들은 부동산 가격의 변동에 관한 전문가들의 예측에 관심을 갖고 있다. 매매가격 혹은 전세가격을 예측하기위해 일반적으로 많이 사용되는 방법은 박스-젠킨스에 기반을 둔 자기회귀이동평균모형이다. 본 논문에서는 자기회귀모형과 다변량 자료분석에서 사용하는 독립성분분석을 결합하여 예측하는 방법을 시도하여 보았다. 매매가격과 전세가격을 두 개의 독립성분으로 재설정하고 독립성분들을 이용하여 예측한 후 역변환을 통해 매매가격과 전세가격을 예측하는 방법을 시도하였다. 그 결과 일반적인 자기회귀이동평균모형을 사용할 때 보다 독립성분을 활용한 예측이 실제 지수에 더 유사한 값들을 얻을 수 있음을 보였다.

• 한국컴퓨터정보학회논문지
• /
• 제17권4호
• /
• pp.59-65
• /
• 2012

고추는 건조과정에서 부패되거나 색이변하는 경우가 발생하므로 건고추 품질을 높이기위해서는 건고추를 선별 할 수 있는 기법이 필요하다. 독립성분분석은 블라인드소스분리에서 가장 널리 사용되는 방법으로 이 기법을 사용하여 건조시킨 고추 영상에서 가장 중요한 성분에 대한 농축영상을 얻는다. 취득한 농축영상은 일반 이진(BW) 영상과 달리 주요 성분만 반영한 것으로 영상의 주요 성분 분포 상태를 알 수 있으며 품질을 판단하여 선별하는 것이 가능하다. 또한, 추출된 농축영상의 크기는 고추의 매운 맛을 내는 주요 성분인 캡사이신류의 양과 관련성이 있음을 알 수 있다. ICA 독립성분을 기반으로 한 농축영상 추출을 통해 고추 건조과정에서 부패되어 색상이 좋지 않거나 캡사이신류과 같은 주요 성분이 없게 된 고추를 선별해하는 방법을 제안한다.

• 전산구조공학
• /
• 제25권3호
• /
• pp.43-48
• /
• 2012

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅳ
• /
• pp.2355-2358
• /
• 2003

The purpose of this paper is to investigate two major feature extraction techniques based on generic modular face recognition system. Detailed algorithms are described for principal component analysis (PCA) and independent component analysis (ICA). PCA and ICA ate statistical techniques for feature extraction and their incorporation into a face recognition system requires numerous design decisions. We explicitly state the design decisions by introducing a modular-based face recognition system since some of these decision are not documented in the literature. We explored different implementations of each module, and evaluate the statistical feature extraction algorithms based on the FERET performance evaluation protocol (the de facto standard method for evaluating face recognition algorithms). In this paper, we perform two experiments. In the first experiment, we report performance results on the FERET database based on PCA. In the second experiment, we examine performance variations based on ICA feature extraction algorithm. The experimental results are reported using four different categories of image sets including front, lighting, and duplicate images.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2002년도 추계학술대회 및 정기총회
• /
• pp.329-332
• /
• 2002

본 논문에서는 효과적인 정보를 표현하는 ICA(Independent Component Analysis)-Factorial 표현 방법을 이용하여 얼굴감정인식을 수행한다. 얼굴감정인식은 두 단계인 특징추출과 인식단계에 의해 이루어진다. 먼저 특징추출방법은 PCA(Principal Component Analysis)을 이용하여 얼굴영상의 고차원 공간을 저차원 특징공간으로 변환한 후 ICA-factorial 표현방법을 통해 좀 더 효과적으로 특징벡터를 추출한다. 인식단계는 최소거리 분류방법인 유클리디안 거리를 이용하여 얼굴감정을 인식한다. 이 방법의 유용성을 설명하기 위해 6개의 기본감정(행복, 슬픔, 화남, 놀람, 공포, 혐오)에 대해 얼굴데이터베이스를 구축하고, 기존의 방법인 Eigenfaces, Fishefaces와 비교하여 좋은 인식성능을 보이고자 한다.

• 한국산업융합학회 논문집
• /
• 제9권4호
• /
• pp.263-268
• /
• 2006

This paper presents a real-time verification system by extracting a features of multimodal biometrics using hybrid method, which is combined the moment balance and the independent component analysis(ICA). The moment balance is applied to reduce the computation loads by extracting the validity signal due to exclude the needless backgrounds of multimodal biometrics. ICA is also applied to increase the verification performance by removing the overlapping signals due to extract the statistically independent basis of signals. Multimodal biometrics are used both the faces and the fingerprints which are acquired by Web camera and acquisition device, respectively. The proposed system has been applied to the fusion problems of 48 faces and 48 fingerprints(24 persons * 2 scenes) of 320*240 pixels, respectively. The experimental results show that the proposed system has a superior verification performances(speed, rate).

• 대한의용생체공학회:의공학회지
• /
• 제26권3호
• /
• pp.177-184
• /
• 2005

In this paper, we proposed a new hearing aid algorithm to improve SNR(signal to noise ratio) of noisy speech signal and speech perception. The proposed hearing aid algorithm is a multi-band loudness compensation based independent component analysis (ICA). The proposed algorithm was compared with a conventional spectral subtraction algorithm on behind-the-ear type hearing aid. The proposed algorithm successfully separated a target speech signal from background noise and from a mixture of the speech signals. The algorithms were compared each other by means of SNR. The average improvement of SNR by ICA based algorithm was 16.64dB, whereas spectral subtraction algorithm was 8.67dB. From the clinical tests, we concluded that our proposed algorithm would help hearing aid user to hear clearly a target speech in noisy conditions.

• 대한조선학회논문집
• /
• 제49권6호
• /
• pp.498-503
• /
• 2012

Independent component analysis (ICA) technique based on statistical independency of the signals is known as suitable to identify the source signals by measuring and separating mixed signals through transfer paths and has successfully applied in the field of medical care, communications and so forth. In this study, the ICA technique is introduced for the identification of excitation sources from measured vibration signals of structures, which can be done by evaluating negentropy of centered and whitened vibration signals and correlation of separated signals. To validate the method, numerical analyses are carried out for a plate and a cylinder structure. The results show that the method can be applied efficiently to source identification of complex structures. Nevertheless, additional studies would be required to complement problems of occasional inaccuracy.