• 한국음향학회지
• /
• 제25권2호
• /
• pp.78-87
• /
• 2006

본 논문에서는 알려지지 않은 잡음 환경에서 강인한 음성 인식 성능을 위하여 missing-feature복구 기법을 다루며, 베이시안 분류기를 기반으로 하는 마스크 예측 기법의 성능을 향상시킬 수 있는 방법을 제안한다. 기존의 마스크 예측 기법에서는 배경 잡음 종류에 독립적인 성능을 위해 전 주파수 대역을 분할하여 발생시킨 유색 잡음을 마스크 예측기의 훈련에 이용하였으나, 제한된 양의 훈련 데이터베이스 조건에서는 성능의 한계가 불가피하다. 보다 다양한 잡음 스펙트럼을 반영하면서 마스크 예측의 성능을 향상시키기 위해, 서로 다른 주파수 대역에 독립적인 구조를 가지는 베이시안 분류기를 제안하며, 훈련에 사용하는 유색 잡음의 생성 방식을 이에 맞게 수정한다. 각각의 주파수 대역을 분할하여 유색 잡음을 생성함으로써 다양한 잡음 환경을 반영하는 동시에 훈련 데이터베이스 부족 문제를 줄일 수 있다. 제안하는 마스크 예측 기법을 클러스터 기반의 missing-feature 복구 기법과 결합하여 음성 인식기에 적용함으로써 성능을 평가한다. 실험 결과는 제안한 기법이 백색 잡음, 자동차잡음, 배경 음악환경에서 기존의 방법에 비해 향상된 성능을 가짐을 입증한다.

• 한국정보통신학회논문지
• /
• 제19권12호
• /
• pp.2844-2851
• /
• 2015

다수의 2차원 객체 영상으로부터 3차원 형상을 복원하는 방법은 컴퓨터 비젼 분야에서 널리 연구되고 있다. 복원된 3차원 형상의 정확도 개선을 위해서는 잡음 영향을 줄이거나 영상 프레임 수를 확보하는 것이 무엇보다 중요하다. 그렇지만 특징점 추정 시 잡음은 잠재적으로 내포되고, 관측행렬을 구성하는 영상 프레임 수는 특징점 추적 실패, 장애요소 또는 낮은 해상력 등에 의해 일반적으로 감소하게 된다. 그래서 잠음 환경 하에 손실된 특징점을 보다 정확히 보정하여 사용 가능한 영상 프레임 수를 확보하는 것이 필수적이다. 따라서 우리는 잡음 분포 하에서 기하학적 특성을 이용해 손실 특징점의 오차 거리와 방향을 직접 제어할 수 있는 분석적 접근방법을 제안한다. 제안한 방법의 우수성은 합성과 실제 객체에 대한 실험 결과를 통해서 검증한다.

• International Journal of Computer Science & Network Security
• /
• 제21권3호
• /
• pp.177-184
• /
• 2021

All over the world, people are affected by many chronic diseases and medical practitioners are working hard to find out the symptoms and remedies for the diseases. Many researchers focus on the feature detection of the disease and trying to get a better health recommendation system. It is necessary to detect the features automatically to provide the most relevant solution for the disease. This research gives the framework of Health Recommendation System (HRS) for identification of relevant and non-redundant features in the dataset for prediction and recommendation of diseases. This system consists of three phases such as Pre-processing, Feature Selection and Performance evaluation. It supports for handling of missing and noisy data using the proposed Imputation of missing data and noise detection based Pre-processing algorithm (IMDNDP). The selection of features from the pre-processed dataset is performed by proposed ensemble-based feature selection using an expert's knowledge (EFS-EK). It is very difficult to detect and monitor the diseases manually and also needs the expertise in the field so that process becomes time consuming. Finally, the prediction and recommendation can be done using Support Vector Machine (SVM) and rule-based approaches.

• 한국정보전자통신기술학회논문지
• /
• 제17권3호
• /
• pp.176-182
• /
• 2024

본 연구는 결측치 비율이 높은 시계열 데이터를 효과적으로 분석하고 예측할 수 있는 머신러닝 모델을 구축하기 위해 다양한 결측치 처리 방법을 비교 분석하였다. 이를 위해 PSMF(Predictive State Model Filtering), MissForest, IBFI(Imputation By Feature Importance) 방법을 적용하였으며, 이후 LightGBM, XGBoost, EBM(Explainable Boosting Machines) 머신러닝 모델을 사용하여 예측 성능을 평가하였다. 연구 결과, 결측치 처리 방법 중에서는 MissForest와 IBFI가 비선형적 데이터 패턴을 잘 반영하여 가장 높은 성능을 나타냈으며, 머신러닝 모델 중에서는 XGBoost와 EBM 모델이 LightGBM 모델보다 더 높은 성능을 보였다. 본 연구는 결측치 비율이 높은 시계열 데이터의 분석 및 예측에 있어 비선형적 결측치 처리 방법과 머신러닝 모델의 조합이 중요함을 강조하며, 실무적으로 유용한 방법론을 제시하였다.

• International Journal of Computer Science & Network Security
• /
• 제21권1호
• /
• pp.220-225
• /
• 2021

Human population growth rate is an important parameter for real-world planning. Common approaches rely upon fixed parameters like human population, mortality rate, fertility rate, which is collected historically to determine the region's population growth rate. Literature does not provide a solution for areas with no historical knowledge. In such areas, machine learning can solve the problem, but a multitude of machine learning algorithm makes it difficult to determine the best approach. Further, the missing feature is a common real-world problem. Thus, it is essential to compare and select the machine learning techniques which provide the best and most robust in the presence of missing features. This study compares 17 machine learning techniques (base learners and ensemble learners) performance in predicting the human population growth rate of the country. Among the 17 machine learning techniques, random forest outperformed all the other techniques both in predictive performance and robustness towards missing features. Thus, the study successfully demonstrates and compares machine learning techniques to predict the human population growth rate in settings where historical data and feature information is not available. Further, the study provides the best machine learning algorithm for performing population growth rate prediction.

• 말소리와 음성과학
• /
• 제6권4호
• /
• pp.127-132
• /
• 2014

This paper describes a robust speech recognition technique by reconstructing spectral components mismatched with a training environment. Although the cluster-based reconstruction method can compensate the unreliable components from reliable components in the same spectral vector by assuming an independent, identically distributed Gaussian-mixture process of training spectral vectors, the presented method exploits the temporal dependency of speech to reconstruct the components by introducing a hidden-Markov-model prior which incorporates an internal state transition plausible for an observed spectral vector sequence. The experimental results indicate that the described method can provide temporally consistent reconstruction and further improve recognition performance on average compared to the conventional method.

• 한국언어정보학회:학술대회논문집
• /
• 한국언어정보학회 2007년도 정기학술대회
• /
• pp.234-240
• /
• 2007

This paper argues that various kinds of displaced structures in English should be licensed by a more explicitly formulated type of rule schema in order to deal with what is called weak connectivity in English. This paper claims that the filler and the gap site cannot maintain the total identity of features but a partial overlap since the two positions need to obey the structural forces that come from occupying respective positions. One such case is the missing object construction where the subject fillers and the object gaps are to observe requirements that are imposed on the respective positions. Others include passive constructions and topicalized structures. In this paper, it is argued that the feature discrepancy comes from the different syntactic positions in which the fillers are assumed to be located before and after displacement. In order to capture this type of mismatch, syntactically relevant features are handled separately from the semantically motivated features in order to deal with the syntactically imposed requirements.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권3호
• /
• pp.881-895
• /
• 2023

Modern image inpainting techniques based on deep learning have achieved remarkable performance, and more and more people are working on repairing more complex and larger missing areas, although this is still challenging, especially for facial image inpainting. For a face image with a huge missing area, there are very few valid pixels available; however, people have an ability to imagine the complete picture in their mind according to their subjective will. It is important to simulate this capability while maintaining the identity features of the face as much as possible. To achieve this goal, we propose a three-stage network model, which we refer to as the identity and structure feature refinement network (ISFRNet). ISFRNet is based on 1) a pre-trained pSp-styleGAN model that generates an extremely realistic face image with rich structural features; 2) a shallow structured network with a small receptive field; and 3) a modified U-net with two encoders and a decoder, which has a large receptive field. We choose structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), L1 Loss and learned perceptual image patch similarity (LPIPS) to evaluate our model. When the missing region is 20%-40%, the above four metric scores of our model are 28.12, 0.942, 0.015 and 0.090, respectively. When the lost area is between 40% and 60%, the metric scores are 23.31, 0.840, 0.053 and 0.177, respectively. Our inpainting network not only guarantees excellent face identity feature recovery but also exhibits state-of-the-art performance compared to other multi-stage refinement models.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2009년도 IWAIT
• /
• pp.714-718
• /
• 2009

3D reconstruction of a human face from an image sequence remains an important problem in computer vision. We propose a method, based on a factorization algorithm, that reconstructs a 3D face model from short image sequences exhibiting rotational motion. Factorization algorithms can recover structure and motion simultaneously from one image sequence, but they usually require that all feature points be well tracked. Under rotational motion, however, feature tracking often fails due to occlusion and frame out of features. Additionally, the paucity of images may make feature tracking more difficult or decrease reconstruction accuracy. The proposed 3D reconstruction approach can handle short image sequences exhibiting rotational motion wherein feature points are likely to be missing. We implement the proposal as a reconstruction method; it employs image sequence division and a feature tracking method that uses Active Appearance Models to avoid the failure of feature tracking. Experiments conducted on an image sequence of a human face demonstrate the effectiveness of the proposed method.

• 한국치위생학회지
• /
• 제17권3호
• /
• pp.331-342
• /
• 2017

Objectives: This study aimed at helping oral health prevention of the blind and related management plan, which is defined as the influence factors between missing and filled permanent teeth index and general feature and oral health behavior of the blind in Korea (estimates 229,678 persons) using data of the 6th Korea National Health and Nutrition Examination Survey from 2014 Korea Centers For Disease Control and Prevention. Methods: The blind over the age of 30 were selected as study subjects who have conducted health survey and dental inspections in KNHANES VI-2. Estimates of the subjects were 229,67 persons. For analyzing data, general linear models: GLM and covariance analysis were conducted to identify the relation between general feature and oral health behavior and missing and filled permanent teeth index. SPSS 21 statistical program was used, which is possible to conduct complex sampling design, and the significance level was 0.05. Results: The missing and filled permanent teeth index was 8.58 points. Regarding the results of the analysis, R-squared of the missing and filled permanent teeth index depending on general features of the blind was 0.839 points, which shows gender, age, residence, education level, individual income, disability rating, kinds of health insurance, marital status and recipient of basic living had an effect on the missing and filled permanent teeth index. R2 of the missing and filled permanent teeth index depending on oral health form of the blind was 0.728 points, which shows oral examination, dental treatment, smoking and toothbrushing after lunch had an effect on the missing and filled permanent teeth index. Conclusions: With the result of this study, we found the oral health actual condition of the blind in Korea. Therefore, it is considered that the government needs to introduce the personalized oral health education program to maintain oral health of the blind and to develop a program that uses braille and voice device which enables to access and utilize to improve oral health behavior that the government could use it as a reference to establish the policy plan.