• International Journal of Advanced Culture Technology
• /
• v.5 no.1
• /
• pp.83-89
• /
• 2017

In this paper, we propose an extended method of one-hot Long Short-Term Memory (LSTM) and evaluate the performance on spam filtering task. Most of traditional methods proposed for spam filtering task use word occurrences to represent spam or non-spam messages and all syntactic and semantic information are ignored. Major issue appears when both spam and non-spam messages share many common words and noise words. Therefore, it becomes challenging to the system to filter correct labels between spam and non-spam. Unlike previous studies on information filtering task, instead of using only word occurrence and word context as in probabilistic models, we apply a neural network-based approach to train the system filter for a better performance. In addition to one-hot representation, using term weight with attention mechanism allows classifier to focus on potential words which most likely appear in spam and non-spam collection. As a result, we obtained some improvement over the performances of the previous methods. We find out using region embedding and pooling features on the top of LSTM along with attention mechanism allows system to explore a better document representation for filtering task in general.

• International Area Studies Review
• /
• v.12 no.3
• /
• pp.195-216
• /
• 2008

It is generally accepted that the momentum strategies are effective in the short-term, and the contrarian strategies are profitable in the long run in major stock markets in the world. In Korean market, however, the contrarian is considered effective investment strategy both in the short- and long-term. We investigate whether this is true after 1999, and try to find out the reasons for this phenomena. We found that the contrarian strategies are still effective. Foreign investors showed consistent investment behavior both in Korean and abroad: they followed momentum in the short-tem, and contrarian in the longer-term. The individual investors, who are thought to be noise trader, showed different behavior. They followed contrarian strategies both in the short-and long-term. The reason that the contrarian is observed in Korean market regardless of the investment horizon is thought to be the irrtional behavior of individual investors.

• Korean Journal of Remote Sensing
• /
• v.26 no.4
• /
• pp.403-410
• /
• 2010

NDVI(Normalized Difference Vegetation Index), which is broadly used as short-term data composite, is an important parameter for climate change and long-term land surface monitoring. Although atmospheric correction is performed, NDVI dramatically appears several low peak noise in the long-term time series. They are related to various contaminated sources, such as cloud masking problem and wet ground condition. This study suggests a simple method through harmonic analysis for reducing NDVI noise using SPOT/VGT NDVI 10-day MVC data. The harmonic analysis method is compared with the polynomial regression method suggested previously. The polynomial regression method overestimates the NDVI values in the time series. The proposed method showed an improvement in NDVI correction of low peak and overestimation.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.1108-1109
• /
• 2023

본 논문은 선박 내부 소음을 효과적으로 감소시키기 위한 ANC(Active Noise Cancellation)및 인공 지능 (AI) 결합 시스템의 개발과 적용에 관한 연구를 다룬다. 선박 환경에서의 소음은 승원의 스트레스 증가와 불편을 초래하므로, 이를 해결하기 위한 방법을 제안하고자 한다. 외부 소음과 내부 소음 데이터를 수집하고, STFT(Short-Time Fourier Transform)알고리즘을 통해 소음 데이터를 분석 가능한 형태로 전처리한다. 그 후, LSTM(Long Short-Term Memory)알고리즘을 사용하여 선박 외부에서 발생한 소음을 입력으로 받아 내부에서 들리는 외부 소음을 예측하고 제어하는 모델을 훈련시킨다. 이후 최적화 과정을 거쳐 예측 소음의 반대 파형을 생성 및 출력을 통해 ANC 를 구현한다.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.12 no.3
• /
• pp.236-242
• /
• 2002

This paper presents some advanced and supplemental methods to enhance the accuracy In case of calculating geometric divergence attenuation, attenuation by multiple screening structures, ground attenuation at unflat surfaces of sound during propagation outdoors by the methods specified in ISO 9613-2. Moreover, a calculation method for considering short-term wind effect, specified in ASJ Model-1998, is also introduced. To verity the accuracy of adopted methods, we have carried out highway traffic noise prediction and measurement at tile twelve locations appearing representative road shapes and structures, such as flat, retained cut, elevated, barrier-constructed roads. From the results, we have confirmed the predicted results show good correspondence with the measured at direct, diffracted and reflected sound fields within 30 m from the center of near side lane.

• Transactions of the Society of Information Storage Systems
• /
• v.3 no.3
• /
• pp.113-117
• /
• 2007

재생층이 상변화물질로 이루어진 초해상 광디스크에 있어서, 기록 전과 후에 thermal curing 을 실시하여 신호품질과 재생안정성이 크게 향상되었다. Pre-curing으로 수천 회 이하에서 나타나는 short-term stability가 향상되었고, post-curing 한 후 mid-term stability 가 향상되었다. 그리고, pre- and post-curing 후 noise level 은 전반적으로 $1{\sim}2dB$가 낮아졌고, CNR은 $2{\sim}3dB$, jitter 는 $2{\sim}3%$ 가 향상되었다.

• 전기의세계
• /
• v.22 no.4
• /
• pp.17-24
• /
• 1973

In the view point of practical engineering the identification problem may be considered as a problem to determine the optimal model in the sense of minimizing a given criterion function using the input-output records of the plant. In the system identification the statistical approach has been known to be very effective when the topological structure of the system and the statistical characteristics of the observation noises are known a priori. But in the practical situation there are many cases when the inforhation about the observation noises or the system noises are not available a priori. Here, the authors propose a new identification method which can be used effectively even in the cases when the variances of observation noises are unknown a priori. In the method, the identification of unknown parameters of a linear diserete system is achieved by minimizing the improved quadratic criterion function which is composed of the term of square equation errors and the term to eliminate the affection of observation noises. The method also gives the estimate of noise variance. Numerical computations for several examples show that the proposed procedure gives satisfactory results even when the short time observation data are provided.

• Journal of the Korean Mathematical Society
• /
• v.59 no.2
• /
• pp.217-233
• /
• 2022

Although machine learning shows state-of-the-art performance in a variety of fields, it is short a theoretical understanding of how machine learning works. Recently, theoretical approaches are actively being studied, and there are results for one of them, margin and its distribution. In this paper, especially we focused on the role of margin in the perturbations of inputs and parameters. We show a generalization bound for two cases, a linear model for binary classification and neural networks for multi-classification, when the inputs have normal distributed random noises. The additional generalization term caused by random noises is related to margin and exponentially inversely proportional to the noise level for binary classification. And in neural networks, the additional generalization term depends on (input dimension) × (norms of input and weights). For these results, we used the PAC-Bayesian framework. This paper is considering random noises and margin together, and it will be helpful to a better understanding of model sensitivity and the construction of robust generalization.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2001.11b
• /
• pp.1280-1286
• /
• 2001

This paper presents some advanced and supplemental methods to enhance the accuracy in case of calculating geometric divergence attenuation, attenuation by multiple screening structures, ground attenuation at unflat surfaces of sound during propagation outdoors by the methods specified in ISO 9613-2. Moreover, a calculation method for considering short-term wind effect, specified in ASJ Model-1998, is also introduced. To verify the accuracy of adopted methods, we have carried out highway traffic noise prediction and measurement at the twelve locations appearing representative road shapes and structures, such as flat, retained cut, elevated, barrier-constructed roads. From the results, we have confirmed the predicted results show good correspondence with the measured at direct, diffracted and reflected sound fields within 30m from the center of near side lane.

• Journal of Astronomy and Space Sciences
• /
• v.28 no.2
• /
• pp.103-108
• /
• 2011

The latitudinal variation of sunspots appearing during the period from 1874 to 2009 has been studied in terms of centerof-latitude (COL). The butterfly diagram has been used to study the evolution of the magnetic field and the dynamics at the bottom of the solar convection zone. Short-term periodicities have been of particular interest, in that they are somehow related to the structure and dynamics of the solar interior. We thus have focused our investigation on shortterm periodicities. We first calculated COL by averaging the latitude of sunspots with the weight function in area. Then, we analyzed the time series of COL using the wavelet transform technique. We found that a periodicity of ~5 years is the most dominant feature in the time series of COL, with the exception of the ~11 year solar cycle itself. This periodicity can be easily understood by considering small humps between the minima in the area-weighted butterfly diagram. However, we find that periodicities of ~1.3 (0.064), ~1.5 (0.056), or ~1.8 (0.046) years ($\frac{1}{month}$), month ), which have been previously suggested as evidence of links between the changing structure of the sunspot zone and the tachocline rotation rate oscillations, are insignificant and inconsistent. We therefore conclude that the only existing short-term periodicity is of ~5 years, and that periodicities of ~1.3, ~1.5, or ~1.8 years are likely to be artifacts due to random noise of small sunspots.