• Title/Summary/Keyword: epoch

Search Result 504, Processing Time 0.029 seconds

Maximum Epoch for Learning Improvement of Second-Order Recurrent Neural Network Inferring Regular Grammars (이차 순환신경망에서 정규문법의 학습을 위한 최대 epoch 결정)

  • 정현기;정순호
    • Journal of Korea Multimedia Society
    • /
    • v.2 no.4
    • /
    • pp.468-475
    • /
    • 1999
  • Learning algorithm of SRNN doesn't use analytic maximum epoch, so that its performance is inefficient and its cost is high. In this paper, with the proper maximum epoch, we improve teaming efficiency. We first describe cost function of maximum epoch and computation time theoretically Then, using it, we propose that maximum epoch must be between 400 and 500. Estimated maximum epoch is verified by experiment.

  • PDF

Physical Activity Assessment of Preschool Children Using Accelerometer - Including Comparison of Reintegrating Counts of Different Epoch Lengths - (가속도계를 이용한 미취학아동의 신체활동 평가 - 자료요약주기의 재통합 측정치와의 비교를 포함하여 -)

  • Kim, Ji-Yeon;Choi, Yeon-Jung;Ju, Mun-Jeong;Kim, Eun-Kyung
    • Journal of the Korean Dietetic Association
    • /
    • v.22 no.3
    • /
    • pp.214-224
    • /
    • 2016
  • The purpose of this study was to assess the physical activity of preschool children using an accelerometer and investigate differences related to epoch length setting during use of the accelerometer. Subjects of the study were 26 children (12 boys and 14 girls) at the age of 5, enrolled in one preschool located in Gangneung. From 9:00 a.m. to 4:00 p.m. (7 hours period), every child wore a total of four accelerometers (ActiGraph $GT3X^+$, USA), including three attached to the left hip (with epoch length set at 5 seconds, 15 seconds, and 30 seconds). For comparison purposes, a forth was attached to the opposite position, with epoch length set at 5 seconds. Data collected using 15s epoch and 30s epoch (single larger epoch) were compared with those obtained after reintegration of 5s to 15s epoch, 5s to 30s epoch, and 15s to 30s epoch, respectively (smaller epochs reintegrated). According to the results of this study, there were no significant differences in VM between 30s epoch and 5s to 30s epoch reintegrated and in MVPA (moderate-to-vigorous physical activity), between 15s epoch and 5s to 15s epoch reintegrated. From the Bland-Altman plot, reintegration of 15s to 30s epoch in VM and reintegrations of 15s to 30s epoch and 5s to 15s epoch in MVPA can be recommended for assessing physical activity in preschool children. Further research is needed into the reintegration method while using an accelerometer for assessment of energy expenditure in children.

A Study on GPS surveying using Epoch-By-Epoch algorithm (Epoch-By-Epoch 알고리즘을 활용한 GPS 측량 실험)

  • 최윤수;고준환;이기도;박지혜
    • Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
    • /
    • 2004.04a
    • /
    • pp.57-57
    • /
    • 2004
  • This study analyzed and compared the results of baseline processing using Epoch-By-Epoch algorithm which is not required initialization compared to conventional kinematic surveying which is required initialization There are rarely differences between 24 hours data of 30 seconds interval and 90 seconds of 30 seconds when it is processed with 26km baseline. This helps with economic surveying using data of GPS CORS

  • PDF

A Study on the Epoch Extraction of Voicd Speech (유성음 구간에서의 Epoch 추출에 관한 연구)

  • 강동규
    • Proceedings of the Acoustical Society of Korea Conference
    • /
    • 1995.06a
    • /
    • pp.216-220
    • /
    • 1995
  • 음성 신호처리분야의 기반을 이루고 잇는 선형예측기법으로 성대폐쇄구간 분석이 가능해질 경우 특히 음성합성가 부호화 시스템의 상당한 성능개선을 기대할 수 있기 때문에 최근 관련 분야에서 높은 신뢰도를 갖는 GCI 검출 알고리즘 개발에 많은 관심을 보이고 있다. 성대폐쇄구간 검출에서 가장 중요한 것은 성대폐쇄시점에 관한 정보이며, 본 논문에서는 이에 대응될 수 있는 정보인 epoch를 음성신호에서 직접 추출할 수 있는 기법을 제안하였다. 제안된 방법은 프레임 단위별 평균 피치를 참조하여 저역통과된 유성음 신호에서 3구간 영교차점별 평균진폭 변동율에 의해 pseudo-epoch를 검출한다. 대역 통과된 유성음 신호를 이용하여 pseudo-epoch 부근에 존재하는 보다 정밀한 실제 epoch을 최종적으로 결정하였다. 제안된 방법은 단계적으로 epoch가 존재할 수 있는 연역을 좁혀 나아가면서 처리하므로 검출오차를 줄일 수 있었고, 시간영역에서 처리되어 계산량이 적으므로 고속 처리가 가능하였다. 성능평가를 위해 처리결과를 EGG 신호와 비교한 결과 약 2샘플 정도의 오차만을 갖는 우수한 성능을 나타내었다.

  • PDF

Constraining the uncertainties in single-epoch virial black hole masses

  • Park, Dae-Seong;Woo, Jong-Hak
    • The Bulletin of The Korean Astronomical Society
    • /
    • v.36 no.1
    • /
    • pp.49.1-49.1
    • /
    • 2011
  • Utilizing single-epoch spectra and the empirical relation between the size of the broad-line region and AGN continuum luminosity, the so-called single-epoch method has been widely used for estimating AGN black hole masses. However, the systematic uncertainties and the potential biases of this method are not well examined. Taking the full advantage of the high-quality homogeneous spectra from the Lick AGN Monitoring Project (LAMP), we investigate in detail the uncertainties of single-epoch mass estimates by comparing with the reverberation-mapping results. We find that the uncertainty due to AGN variability is less than 0.1 dex, while there is a systematic offset between single-epoch masses and reverberation masses. Particularly, narrow-line Seyfert 1 galaxies show that the Hbeta line widths measured from single-epoch (or mean) spectra are systematically larger than those from rms spectra, indicating a potential bias of single-epoch masses. We will present the detailed measurement method, the test of virial assumption, and the systematic uncertainties.

  • PDF

Characteristics of the Float Solutions in Epoch-by-epoch GNSS Precise Positioning (Epoch-by-epoch GNSS 정밀 측위에서 실수해의 특성)

  • Han, Deok-Hee;Park, Chan-Sik;Hwang, Dong-Hwan;Lee, Sang-Jeong
    • Journal of Institute of Control, Robotics and Systems
    • /
    • v.15 no.12
    • /
    • pp.1254-1257
    • /
    • 2009
  • This paper describes characteristics of the float solutions of epoch-by-epoch GNSS precise positioning. It is shown that the accuracy of the floating solutions which are obtained using both code and carrier phase measurements are same as that of the code measurements only solution. It implies that to get more accurate floating position, the quality of code measurements are more important than that of carrier phase measurements. The results are expanded to multiple frequencies applications to get the same conclusions.

Basic Study for Ambiguity Resolution using Epoch-By-Epoch Algorithm (Epoch-By-Epoch 방법에 의한 미지정수 해결을 위한 기초연구)

  • 박정현;강준묵
    • Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
    • /
    • 2004.11a
    • /
    • pp.95-98
    • /
    • 2004
  • GPS double difference, carrier phase measurements are ambiguous by an unknown inter number of cycle. High precision, relative GPS positioning can be obtained from a short time span of data if the integer double difference ambiguities can be determined efficiently. In this study, we used a ambiguities resolution method indicated by Clyde Goad. And I compared with ambiguities using Lambda method. The difference of ambiguities between two method is below ${\pm}$1.

  • PDF

A comparative study on algorithms for GPS baseline processing (GPS 기선해석을 위한 알고리즘 비교연구)

  • Lee, Ki-Do;Choi, Yun-Soo;Lee, Im-Pyeong
    • 한국지형공간정보학회:학술대회논문집
    • /
    • 2004.10a
    • /
    • pp.39-43
    • /
    • 2004
  • GPS 기선해석에서 가장 중요한 과정은 정수 바이어스를 결정하는 것이다. 기존에 Multi Epoch를 사용한 방법이 일반적이었으나 Single Epoch를 이용한 방법이 개발되었다. 본 연구는 Multi Epoch 및 Single Epoch 알고리듬이 구현된 소프트웨어를 활용하여 여러 가지 데이터 처리 실험을 수행하고 그 결과를 분석하였다. 이를 통해 실시간 정밀 측량에 대한 이용 가능성을 검증하였다.

  • PDF

Tone Jamming Effect on the Epoch Synchronization (Epoch 동기에 대한 Tone Jamming의 영향 분석)

  • 문장욱;은유창;송홍엽;이광억;최영균
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.26 no.9A
    • /
    • pp.1609-1613
    • /
    • 2001
  • 본 논문에서는 실시간 자기 상관 관계를 이용하여 주파수 도약 시점을 찾아내는 epoch 추정기를 소개하고 이것의 성능을 tone jamming 환경 하에서 분석해 보았다. 주파수 도약 방법을 이용하는 시스템은 주로 군 관련 시스템이므로 jamming 환경 하에서의 성능이 매우 중요하다. Jamming이 있을 경우 epoch 추정기는 실제 값보다 작은 값을 나타내는 것을 알 수 있었다. 분석 과정을 수학적으로 유도하였고, 컴퓨터 실험으로 결과를 확인하였다.

  • PDF

SOC Prediction of Lithium-ion Batteries Using LSTM Model

  • Sang-Hyun Lee
    • International Journal of Advanced Culture Technology
    • /
    • v.12 no.3
    • /
    • pp.466-470
    • /
    • 2024
  • This study proposes a deep learning-based LSTM model to predict the state of charge (SOC) of lithium-ion batteries. The model was trained using data collected under various temperature and load conditions, including measurement data from the CS2 lithium-ion battery provided by the University of Maryland College of Engineering. The LSTM model effectively models temporal patterns in the data by learning long-term dependencies. Performance evaluation by epoch showed that the predicted SOC improved from 14.8400 at epoch 10 to 12.4968 at epoch 60, approaching the actual SOC value of 13.5441. The mean absolute error (MAE) and root mean squared error (RMSE) also decreased from 0.9185 and 1.3009 at epoch 10 to 0.2333 and 0.5682 at epoch 60, respectively, indicating continuous improvement in predictive performance. This study demonstrates the validity of the LSTM model for predicting the SOC of lithium-ion batteries and its potential to enhance battery management systems.