• 한국음향학회지
• /
• 제40권2호
• /
• pp.130-138
• /
• 2021

본 논문에서는 지진 이벤트 분류를 위한 다양한 정규화 기법 분석 및 효과적인 합성곱 신경망(Convolutional Neural Network, CNN)기반의 네트워크 구조를 제안하였다. 정규화 기법은 신경망의 학습 속도를 개선할 뿐만 아니라 잡음에 강인한 모습을 보여 준다. 본 논문에서는 지진 이벤트 분류를 위한 딥러닝 모델에서 입력 정규화 및 은닉 레이어 정규화가 모델에 미치는 영향을 분석하였다. 또한, 적용 은닉 레이어의 구조에 따른 다양한 실험을 통해 효과적인 모델을 도출하였다. 다양한 모의실험 결과 입력 데이터 정규화 및 제1 은닉 레이어에 가중치 정규화를 적용한 모델이 가장 안정적인 성능 향상을 보여 주었다.

• Earthquakes and Structures
• /
• 제2권2호
• /
• pp.171-187
• /
• 2011

Rational scaling of design earthquake ground motions for tall buildings is essential for safer, risk-based design of tall buildings. This paper provides the structural designers with an insight for more rational scaling based on drift and input energy demands. Since a resonant sinusoidal motion can be an approximate critical excitation to elastic and inelastic structures under the constraint of acceleration or velocity power, a resonant sinusoidal motion with variable period and duration is used as an input wave of the near-field and far-field ground motions. This enables one to understand clearly the relation of the intensity normalization index of ground motion (maximum acceleration, maximum velocity, acceleration power, velocity power) with the response performance (peak interstory drift, total input energy). It is proved that, when the maximum ground velocity is adopted as the normalization index, the maximum interstory drift exhibits a stable property irrespective of the number of stories. It is further shown that, when the velocity power is adopted as the normalization index, the total input energy exhibits a stable property irrespective of the number of stories. It is finally concluded that the former property on peak drift can hold for the practical design response spectrum-compatible ground motions.

• 한국통신학회논문지
• /
• 제39A권3호
• /
• pp.155-157
• /
• 2014

In this paper, a novel codebook generation strategy is proposed. With the given codebooks, two codeword selection procedures are proposed and analyzed for generating the quantized multiple-input multiple-output (MIMO) channel state information (CSI). Furthermore, three different quantization and normalization strategies are analyzed. The simulation results suggest that the proposed 'quantized channel generation method 2' is the best strategy to reduce the quantization and normalization errors to generate the final quantized MIMO CSI.

• 한국산학기술학회논문지
• /
• 제21권6호
• /
• pp.374-382
• /
• 2020

본 연구에서 저장탄약 신뢰성평가(ASRP: Ammunition Stockpile Reliability Program)의 데이터 특성을 고려하여 입력변수를 줄이는 정규화기법을 제안함으로써 분류성능의 저하 없이 저장탄약 신뢰성분류 인경신경망모델의 학습 속도향상을 목표로 하였다. 탄약의 성능에 대한 기준은 국방규격(KDS: Korea Defense Specification)과 저장탄약 시험절차서(ASTP: Ammunition Stockpile reliability Test Procedure)에 규정되어 있으며, 평가결과 데이터는 이산형과 연속형 데이터가 복합적으로 구성되어 있다. 이러한 저장탄약 신뢰성평가의 데이터 특성을 고려하여 입력변수는 로트 추정 불량률(estimated lot percent nonconforming) 또는 고장률로 정규화 하였다. 또한 입력변수의 unitary hypercube를 유지하기 위하여 최소-최대 정규화를 2차로 수행하는 2단계 정규화 기법을 제안하였다. 제안된 2단계 정규화 기법은 저장탄약 신뢰성평가 데이터를 이용하여 비교한 결과 최소-최대 정규화와 유사하게 AUC(Area Under the ROC Curve)는 0.95 이상이었으며 학습속도는 학습 데이터 수와 은닉 계층의 노드 수에 따라 1.74 ~ 1.99 배 향상되었다.

• 한국통신학회논문지
• /
• 제31권8C호
• /
• pp.801-805
• /
• 2006

A normalization approach to coefficient adaptation in the escalator(ESC) filter structure that conventionally employs least mean square(LMS) algorithm is introduced. Using Taylor's expansion of the local error signal, a normalized form of the ESC-LMS algorithm is derived. Compared with the computational complexity of the conventional ESC-LMS algorithm employs input power estimation for time-varying convergence coefficient using a single-pole low-pass filter, the computational complexity of the proposed method can be reduced by 50% without performance degradation.

• 전자공학회논문지B
• /
• 제30B권9호
• /
• pp.84-93
• /
• 1993

Recently, several nonlinear shape normalization methods have been proposed in order to compensate for the shape distortions in handwritten characters. In this paper, we review these nonlinear shape normalization methods from the two points of view : feature projection and feature density equalization. The former makes feature projection histogram by projecting a certain feature at each point of input image into horizontal-or vertical-axis and the latter equalizes the feature densities of input image by re-sampling the feature projection histogram. A systematic comparison of these methods has been made based on the following criteria: recognition rate, processing speed, computational complexity and measure of variation. Then, we present the result of quantitative evaluation of each method based on these criteria for a large variety of handwritten Hangul syllables.

• 인터넷정보학회논문지
• /
• 제4권3호
• /
• pp.9-14
• /
• 2003

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.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 1993년도 학술논문발표회 논문집 제12권 1호
• /
• pp.238-241
• /
• 1993

We present a new method to normalize spectral variations of different speakers based on physiological studies of hearing. The proposed method uses the cochlear frequency map to warp the input speech spectra by interpolation or decimation. Using this normalization method, we can obtain much improved recognition results for speaker independent speech recognition.

• 한국멀티미디어학회논문지
• /
• 제23권8호
• /
• pp.977-985
• /
• 2020

Aiming at the problem of complex feature extraction and low accuracy in human action recognition, this paper proposed a network structure combining batch normalization algorithm with GoogLeNet network model. Applying Batch Normalization idea in the field of image classification to action recognition field, it improved the algorithm by normalizing the network input training sample by mini-batch. For convolutional network, RGB image was the spatial input, and stacked optical flows was the temporal input. Then, it fused the spatio-temporal networks to get the final action recognition result. It trained and evaluated the architecture on the standard video actions benchmarks of UCF101 and HMDB51, which achieved the accuracy of 93.42% and 67.82%. The results show that the improved convolutional neural network has a significant improvement in improving the recognition rate and has obvious advantages in action recognition.

• 한국통신학회논문지
• /
• 제21권5호
• /
• pp.1345-1359
• /
• 1996

This paper presents a floating point arithmetic unit (FPAU) for supescalar microprocessor that executes fifteen operations such as addition, subtraction, data format converting, and compare operation using two pipelined arithmetic paths and new rounding and normalization scheme. By using two pipelined arithmetic paths, each aritchmetic operation can be assigned into appropriate arithmetic path which high speed operation is possible. The proposed normalization an rouding scheme enables the FPAU to execute roundig operation in parallel with normalization and to reduce timing delay of post-normalization. And by predicting leading one position of results using input operands, leading one detection(LOD) operation to normalize results in the conventional arithmetic unit can be eliminated. Because the FPAU can execuate fifteen single-precision or double-precision floating-point arithmetic operations through three-stage pipelined datapath and support IEEE standard 754, it has appropriate structure which can be ingegrated into superscalar microprocessor.