• 한국음향학회지
• /
• 제23권3호
• /
• pp.253-261
• /
• 2004

본 논문에서는 음성 신호가 지니고 있는 화자 의존적 특징 변수를 변환 시키는 음성 개성 변환 기법이 새롭게 제안되었다. 제안된 방법은 성도 전달 함수의 특성을 반영하는 켑스트럼 벡터와 여기 신호의 특성을 반영하는 피치 값을 변환 대상 변수로 삼았으며, 이들에 대한 변환 기법으로 다중 응답 분류 회귀 트리를 사용하였다. 다중 응답 분류 회귀 트리는 기존의 분류 회귀 트리를 다차원 확장시킨 형태로서, 반응값이 벡터 형태로 존재하는 분류 회귀 트리를 의미한다. 본 논문에서는 기존의 코드북 메핑 방법과 비교하여 제안된 기법의 성능을 평가하였으며, 분류 회귀 트리에 입력되는 관찰값을 다양하게 변화시켜 트리의 복잡도와 변환 성능을 정량적으로 분석하였다. 네 명의 화자를 이용한 음성 개성 변환 실험에서, 기존의 코드북 메핑과 비교하여 객관적으로 우수한 성능을 나타내었으며, 청취 테스트에서도 변환음이 목표로 하는 화자의 음성과 유사함을 관찰할 수 있었다.

• 복식문화연구
• /
• 제12권3호
• /
• pp.339-353
• /
• 2004

Purposes of this study were to analyzed coat types and characteristics of coat of young persons, and search whether fashion trend is reflected on coat. Data collected pictures that they are wearing dress in street of Daejeon city 3 places that there are much the rising generations at November, 1999. This study target was from teens latter half to 20 opening part, 154 women. Data analyzed content analysis, frequency analysis, crossing and the result is as following. First, classification standard of coat was textile fabric, form of detail and ornament. Second, coat could classify in three types, type 1 was traditional duffle coat style that is distinguished by form of detail and ornament(hood and button). Type 2 was classified property of textile fabric that used leather, padding, fur etc., and type 3 was classified by collar detail of woolen fabric coat. Specially, ornamental fur of woolen coat perceived visually strong. And design detail of coat showed significant difference in coat type. That is, duffle coat type was designed patch pocket and toggle, woolen fabric coat type was hidden button and seam pocket. Third, fashion tendency of coat was proved that is reflecting part of predicted fashion trend.

• 대한임베디드공학회논문지
• /
• 제12권3호
• /
• pp.149-157
• /
• 2017

In this paper, we propose BCMR (Block Classification with Monitor and Restriction) to ensure the isolation and to reduce the interference of blocks between a garbage collection and a wear leveling. The proposed BCMR monitors an endurance variation of blocks during the garbage collection and detects hot blocks by making a restriction condition based on this information. The proposal induces a block classification by its update frequency for the garbage collection and the wear leveling, so we will get a prolonged lifetime of NAND flash memory systems. In a performance evaluation, BCMR prolonged the lifetime of NAND flash memory systems by 3.95%, on average and reduced a standard deviation per block by 7.4%, on average.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2012년도 추계학술대회 논문집
• /
• pp.394-399
• /
• 2012

최근 MEMS 센서는 기계상태감시에 있어서 전력소모, 크기, 비용, 이동성, 응용 등에 있어서 각광을 받고 있다. 특히, MEMS 센서는 스마트센서와 통합가능하고, 대량생산이 가능하여 가격이 저렴하다는 장점이 있다. 이와 관련한 기계상태감시를 위한 많은 실험적 연구가 수행되고 있다. 이 논문은 MEMS 센서들을 3 가지 인공지능 분류기 성능평가를 위한 비교연구에 대해 설명하고 있다. 회전기계에 MEMS 가속도와 전류센서들을 부착하여 데이터를 취득했고, 특징추출과 파라미터 최적화를 위해 Cross validation 기법을 사용하였다. MEMS 센서를 이용한 결함분류기 적용은 적합하다고 판단된다.

• 산업경영시스템학회지
• /
• 제43권2호
• /
• pp.33-38
• /
• 2020

Anomaly detection of Machine Learning such as PCA anomaly detection and CNN image classification has been focused on cross-sectional data. In this paper, two approaches has been suggested to apply ML techniques for identifying the failure time of big time series data. PCA anomaly detection to identify time rows as normal or abnormal was suggested by converting subjects identification problem to time domain. CNN image classification was suggested to identify the failure time by re-structuring of time series data, which computed the correlation matrix of one minute data and converted to tiff image format. Also, LASSO, one of feature selection methods, was applied to select the most affecting variables which could identify the failure status. For the empirical study, time series data was collected in seconds from a power generator of 214 components for 25 minutes including 20 minutes before the failure time. The failure time was predicted and detected 9 minutes 17 seconds before the failure time by PCA anomaly detection, but was not detected by the combination of LASSO and PCA because the target variable was binary variable which was assigned on the base of the failure time. CNN image classification with the train data of 10 normal status image and 5 failure status images detected just one minute before.

• 대한원격탐사학회지
• /
• 제25권6호
• /
• pp.531-546
• /
• 2009

The main objective of this study is to analyse the polarimetric characteristics of the various terrain targets by ground-based polarimetric SAR system and to confirm the compatible and effective polarimetric analysis method to reveal the polarization properties of different terrain targets by the GB-SAR. The fully polarimetric GB-SAR data with HH, HV, VH, and VV components were focused using the Deramp-FFT (DF) algorithm. The focused GB-SAR images were processed by the H/A/$\alpha$ polarimetric decomposition and the combined H/$\alpha$ or H/A/$\alpha$ and Wishart classification method. The segmented image and distribution graphs in H/$\alpha$ plane using Cloude and Pottier's method showed a reliable result that this quad-polarization GB-SAR data could be useful to classified corresponding scattering mechanism. The H/$\alpha$-Wishart and H/A/$\alpha$-Wishart classification results showed that a natural media and an artificial target were discriminated by the combined classification, in particular, after applying multi-looking and the Lee refined speckle filter.

• Journal of Information Processing Systems
• /
• 제15권3호
• /
• pp.694-706
• /
• 2019

In this study, we applied the long short-term memory (LSTM) model to classify the cryptocurrency price time series. We collected historic cryptocurrency price time series data and preprocessed them in order to make them clean for use as train and target data. After such preprocessing, the price time series data were systematically encoded into the three-dimensional price tensor representing the past price changes of cryptocurrencies. We also presented our LSTM model structure as well as how to use such price tensor as input data of the LSTM model. In particular, a grid search-based k-fold cross-validation technique was applied to find the most suitable LSTM model parameters. Lastly, through the comparison of the f1-score values, our study showed that the LSTM model outperforms the gradient boosting model, a general machine learning model known to have relatively good prediction performance, for the time series classification of the cryptocurrency price trend. With the LSTM model, we got a performance improvement of about 7% compared to using the GB model.

• 스마트미디어저널
• /
• 제11권11호
• /
• pp.40-48
• /
• 2022

본 논문에서는 인공지능교육의 어려움을 해결하기 위하여 인공지능 교육에 활용이 가능한 교육용 키트를 개발하였다. 이를 통하여 이론 중심에서 실무 위주의 경험을 학습하기 위한 CNN과 OpenCV를 이용하여 컴퓨터 비전 기술을 이용한 사람 인식(Object Detection and Person Detection in Computer Vision)과 특정 오브젝트를 학습시키고 인식시키는 사용자 이미지인식(Your Own Image Recognition), 사용자 객체 분류(Segmentation) 및 세분화(Classification Datasets), 학습된 타켓을 공격하는 IoT하드웨어 제어와 인공지능보드인 Jetson Nano GPIO를 제어함으로써 효과적인 인공지능 학습에 도움이 되는 교재를 개발하여 활용할 수 있도록 하였다.

• 농촌계획
• /
• 제30권2호
• /
• pp.25-34
• /
• 2024

This study is a basic study for reorganizing the facility system of agriculture and rural areas necessary for establishing a rural spatial plan. Accordingly, the newly implemented rural spatial planning system was briefly reviewed. As the scope of the study, the facility-related laws and the classification and classification system of facilities of previous studies were set as the scope of the study. In order to reorganize the facility system in rural areas necessary for establishing a rural space plan, this study compared and analyzed the facilities according to the laws related to the facilities and the use of previous studies. As a result of analyzing 21 target sites for rural agreements with 12 sectors of service facilities in rural areas as indicators, 14 facilities in 8 sectors were found to be commonly introduced for the establishment of living areas in rural areas or regional development. However, the classification of production space facilities related to agriculture as functional facilities necessary for rural life was insufficient. Accordingly, when considering the specificity of rural areas, it is necessary to classify facilities of living spaces in rural areas and production space of agriculture according to their use.

• 센서학회지
• /
• 제33권3호
• /
• pp.139-146
• /
• 2024

Surface-enhanced Raman scattering (SERS) enables the detection of various types of π-conjugated biological and chemical molecules owing to its exceptional sensitivity in obtaining unique spectra, offering nondestructive classification capabilities for target analytes. Herein, we demonstrate an innovative strategy that provides significant machine learning (ML)-enabled predictive SERS platforms through surface-engineered graphene via complementary hybridization with Au nanoparticles (NPs). The hybridized Au NPs/graphene SERS platforms showed exceptional sensitivity (10^-7 M) due to the collaborative strong correlation between the localized electromagnetic effect and the enhanced chemical bonding reactivity. The chemical and physical properties of the demonstrated SERS platform were systematically investigated using microscopy and spectroscopic analysis. Furthermore, an innovative strategy employing ML is proposed to predict various analytes based on a featured Raman spectral database. Using a customized data-preprocessing algorithm, the feature data for ML were extracted from the Raman peak characteristic information, such as intensity, position, and width, from the SERS spectrum data. Additionally, sophisticated evaluations of various types of ML classification models were conducted using k-fold cross-validation (k = 5), showing 99% prediction accuracy.