• 대한수학회지
• /
• 제59권2호
• /
• 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.

• 대한수학회지
• /
• 제52권3호
• /
• pp.447-467
• /
• 2015

In this paper, we consider a multi-risk model based on the policy entrance process with n independent policies. For each policy, the entrance process of the customer is a non-homogeneous Poisson process, and the claim process is a renewal process. The loss process of the single-risk model is a random sum of stochastic processes, and the actual individual claim sizes are described as extended upper negatively dependent (EUND) structure with heavy tails. We derive precise large deviations for the loss process of the multi-risk model after giving the precise large deviations of the single-risk model. Our results extend and improve the existing results in significant ways.

• 대한산업공학회지
• /
• 제31권2호
• /
• pp.164-172
• /
• 2005

A loss function approach to a multiresponse problem is considered, when process parameters are regarded as random variables. The variation of each response may be amplified through so called propagation of error (POE), which is defined as the standard deviation of the transmitted variability in the response as a function of process parameters. The forms of POE for each response and for a pair of responses are proposed and they are reflected in our loss function approach to determine the optimal condition. The proposed method is illustrated using a polymer case. The result is compared with the case where parameter fluctuation is not considered.

• 한국항행학회논문지
• /
• 제14권3호
• /
• pp.358-366
• /
• 2010

무선 네트워크상에서 혼잡과 지연은 네트워크 내에 존재하는 패킷의 수가 과도하게 증가하거나 송신측과 수신측의 전송 균형이 일치되지 않을 때 주로 발생한다. 이러한 혼잡과 지연은 패킷 손실 (pack loss)의 원인이 되며, 패킷손실은 멀티미디어 스트리밍의 성능을 떨어뜨릴 뿐만 아니라 오버헤드를 증가시킨다. 본 논문에서는 무선네트워크의 패킷 손실을 최적화하고 멀티미디어 스트리밍의 QoS 향상을 위한 전송율 기반의 멀티미디어 스트리밍의 최적화 메카니즘을 제안한다. 제안된 기법은 전송율 모니터링과 오버헤드 모니터링에 기반하여 최적화를 수행한다. 이러한 목적을 위하여 본 논문에서는 소스율 제어에 의한 최적화를 수행하도록 하며, 이것은 혼잡, 지연 등의 이슈들을 최적화하기 위한 것이다. 성능 평가는 RED (Random Early Detection), TFRC (TCP-friendly Rate Control) 그리고 제안된 기법으로 수행하였으며, 시뮬레이션 결과 제안된 기법이 RED, TFRC기법에 비해서 패킷손실율, 처리율, 평균 응답율이 보다 효율적임을 알게 되었다.

• 한국통신학회논문지
• /
• 제38B권7호
• /
• pp.553-563
• /
• 2013

IEEE 802.11 무선 랜 기반 방송 서비스를 제공 할 때 한정적인 무선 자원을 이용하여 다수의 사용자들에게 원활하게 서비스하기 위해 유니캐스트 패킷 대신 방송 패킷을 이용한다. 방송 패킷은 일정한 대역폭을 사용하여 다수의 사용자들에게 동시에 패킷을 전송 할 수 있지만 손실 복원이 어려운 단점 있기 때문에 손실 특성 분석을 통한 효율적인 패킷 복원 방안이 요구 된다. 손실의 특성 중에서 일정 구간에서 다수의 패킷이 연속적으로 손실 되는 구간이 있는데 이를 버스트 손실 구간이라고 한다. 평균 패킷 손실율을 가지더라도 랜덤 손실과 버스트 손실의 특성에 따라 구간별 손실에 차이가 발생하기 때문에 같은 손실 복원 기법을 적용하더라도 복원율의 차이가 발생한다. 따라서 손실의 본질을 분석 하고 이를 고려한 손실 복원 방안에 대한 연구가 필요하다. 본 논문에서는 전송률에 따른 Wi-Fi 방송 실험을 통해 생성된 실제 손실 패턴을 바탕으로 4-상태 마코프 모델을 이용하여 버스트 손실의 특성을 분석 하였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2005년도 제36회 하계학술대회 논문집 A
• /
• pp.580-582
• /
• 2005

This paper presents a new method which applies a genetic algorithm(GA) for determining which sectionalizing switch to operate in order to solve the distribution system loss minimization re-configuration problem. The distribution system loss minimization re-configuration problem is in essence a 0-1 planning problem which means that for typical system scales the number of combinations requiring searches becomes extremely large. In order to deal with this problem, a new a roach which applies a GA was presented. Briefly, GA are a type of random number search method, however, they incorporate a multi-point search feature. Further, every point is not is not separately and respectively renewed, therefore, if parallel processing is applied, we can expect a fast solution algorithm to result.

• 한국소음진동공학회논문집
• /
• 제17권11호
• /
• pp.1013-1020
• /
• 2007

In this paper, the predictions and measurements of sound transmission loss(STL) are discussed for various types of acoustical materials and carpets. Random incidence sound transmission losses are measured by the sound intensity method. The in-house software HONUS2005 is used to predict TL and estimate the various physical properties such as the flow resistivity, the structure factor, the porosity, the Possion's ratio, and etc. After this estimation, various multi-layered materials with a steel plate are measured and predicted. In particular, Carpets are assumed to be membranes to predict acoustical performance. To confirm this assumption, double and triple-layered cases are also observed including two different kinds of carpets.

• 한국시뮬레이션학회논문지
• /
• 제8권4호
• /
• pp.1-8
• /
• 1999

This paper proposes an ATM switch architecture called Output Queued Batcher-Banyan switch (OQBBS). It consists of a Sorting Module, Expanding Module, and Output Queueing Modules. The principles of channel grouping and output queueing are used to increase the maximum throughput of an ATM switch. One distinctive feature of the OQBBS is that multiple cells can be simultaneously delivered to their desired output. The switch architecture is shown to be modular and easily expandable. The performance of the OQBBS in terms of throughput, cell delays, and cell loss rate under uniform random traffic condition is evaluated by computer simulation. The throughput and the average cell delay are close to the ideal performance behavior of a fully connected output queued crossbar switch. It is also shown that the OQBBS meets the cell loss probability requirement of $10^{-6}$.

• Communications for Statistical Applications and Methods
• /
• 제18권2호
• /
• pp.165-170
• /
• 2011

Support vector quantile regression(SVQR) is capable of providing a good description of the linear and nonlinear relationships among random variables. In this paper we propose a sparse SVQR to overcome a limitation of SVQR, nonsparsity. The asymmetric e-insensitive loss function is used to efficiently provide sparsity. The experimental results are presented to illustrate the performance of the proposed method by comparing it with nonsparse SVQR.

• Communications for Statistical Applications and Methods
• /
• 제17권2호
• /
• pp.183-191
• /
• 2010

Support vector quantile regression(SVQR) is capable of providing more complete description of the linear and nonlinear relationships among random variables. In this paper we propose an iterative reweighted least squares(IRWLS) procedure to solve the problem of SVQR with a weighted quadratic loss function. Furthermore, we introduce the generalized approximate cross validation function to select the hyperparameters which affect the performance of SVQR. Experimental results are then presented which illustrate the performance of the IRWLS procedure for SVQR.