• 대한수학회지
• /
• 제54권4호
• /
• pp.1317-1329
• /
• 2017

A Boolean rank one matrix can be factored as $\text{uv}^t$ for vectors u and v of appropriate orders. The perimeter of this Boolean rank one matrix is the number of nonzero entries in u plus the number of nonzero entries in v. A Boolean matrix of Boolean rank k is the sum of k Boolean rank one matrices, a rank one decomposition. The perimeter of a Boolean matrix A of Boolean rank k is the minimum over all Boolean rank one decompositions of A of the sums of perimeters of the Boolean rank one matrices. The arctic rank of a Boolean matrix is one half the perimeter. In this article we characterize the linear operators that preserve the symmetric arctic rank of symmetric Boolean matrices.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2010년도 하계학술대회
• /
• pp.9-11
• /
• 2010

Multiple-input multiple-output (MIMO) systems assisted by multi-relays with single antenna are considered. Signal transmission consists of two hops. In the first hop, the source node broadcasts the vector symbols to all relays, then all relays forward the received signals multiplied by each power gain to the destination simultaneously. Unlike the case of full cooperation between relays such as single relay with multiple antennas, in our case there is no closed form solution for optimal relay power gain with respect to minimum mean square error (MMSE). Thus we propose an alternative approach in which we use an approximation of the cost function based on rank-one matrix decomposition. As a cost function, we choose the trace of MSE matrix. We give several simulation results to validate that our proposed method obtains a negligible performance loss compared to optimal solution obtained by exhaustive search.

• 한국정보과학회논문지:데이타베이스
• /
• 제33권7호
• /
• pp.708-714
• /
• 2006

웹은 정보의 저장 및 검색에 있어서 보편적인 매체가 되고 있다. 웹에서 정보 검색은 검색엔진을 출발점으로 이용하는 것이 대부분이지만, 그 결과는 사용자의 요구와 늘 일치하는 것은 아니며 때로는 의도적으로 조작된 검색 결과가 제시되기도 한다. 검색엔진의 데이타를 의도적으로 조작하는 것을 스패밍(spamming)이라고 부르며, 다양한 스패밍과 방지기술이 있지만, 최근에 각광을 받고있는 링크기반 검색 방식에는 스패밍이 쉽지 않은 것으로 알려져 있다. 그러나 이러한 방식에서도 구글폭탄(Google Bombing)과 같이 페이지점수법(PageRank)을 조작할 수 있는 약점이 있다. 본 논문에서는 이러한 약점을 방지할 수 있는 알고리즘을 제시한다. 기본적으로 링크 기반 검색 방식을 기초로 하여 웹을 하나의 유향 레이블 그래프로 인식하여 각 웹 페이지들은 하나의 노드로, 하이퍼링크는 에지로 표현함에 있어서 본 연구에서는 링크구조를 기반으로 링크내역(link context)을 부여하고 이를 에지의 레이블로 사용한다. 링크내역과 대상 페이지 사이의 유사도를 구하고, 이것을 이용하여 페이지점수법의 인접행렬을 재구성하는 방법을 취했다. 결과로써 기존의 방법 및 특이값 추출기법(SVD)에 기반한 새로운 기준을 도입해 그 효과를 입증했다.

• 대한수학회지
• /
• 제57권2호
• /
• pp.401-414
• /
• 2020

It is computationally expensive to compute principal components from scratch at every update or downdate when new data arrive and existing data are truncated from the data matrix frequently. To overcome this limitations, incremental principal component analysis is considered. Specifically, we present a sliding window based efficient incremental principal component computation from a covariance matrix which comprises of two procedures; simultaneous update and downdate of principal components, followed by the rank-one matrix update. Additionally we track the accurate decomposition error and the adaptive numerical rank. Experiments show that the proposed algorithm enables a faster execution speed and no-meaningful decomposition error differences compared to typical incremental principal component analysis algorithms, thereby maintaining a good approximation for the principal components.

• 방송공학회논문지
• /
• 제26권2호
• /
• pp.125-131
• /
• 2021

최근 CNN(Convolutional Neural Network)은 영상 분류, 객체 인식, 화질 개선 등 다양한 비전 분야에서 우수한 성능을 보여주고 있다. 그러나 많은 메모리와 계산량이 요구되어 모바일 또는 IoT(Internet of Things) 장치와 같은 저전력 디바이스에 적용하기에는 제한이 따른다. 이에, CNN 모델의 임무 성능을 유지하면서 네트워크 모델을 압축하는 연구가 진행되고 있다. 본 논문에서는 행렬 분해 기술인 저계수 행렬 근사(Low-rank approximation)와 CP(Canonical Polyadic) 분해 기법을 결합한 CNN 모델 압축 기법을 제안한다. 제안기법은 하나의 행렬 분해 기법만을 적용하는 기존의 기법과 달리 CNN의 계층 유형에 따라 두 가지 분해 기법을 선택적으로 적용하여 압축 성능을 높인다. 제안기법의 성능 검증을 위하여 영상 분류 CNN 모델인 VGG-16, ResNet50, 그리고 MobileNetV2 모델을 압축하였고, 계층 유형에 따라 두 가지의 분해 기법을 선택적으로 적용함으로써 저계수 행렬 근사 기법만 적용한 경우 보다 1.5 ~ 12.1 배의 동일한 압축률에서 분류 성능이 향상됨을 확인하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권6호
• /
• pp.2634-2648
• /
• 2020

In this paper, an adaptive sparse singular value decomposition (ASSVD) algorithm is proposed to estimate the signal matrix when only one data matrix is observed and there is high dimensional white noise, in which we assume that the signal matrix is low-rank and has sparse singular vectors, i.e. it is a simultaneously low-rank and sparse matrix. It is a structured matrix since the non-zero entries are confined on some small blocks. The proposed algorithm estimates the singular values and vectors separable by exploring the structure of singular vectors, in which the recent developments in Random Matrix Theory known as anisotropic Marchenko-Pastur law are used. And then we prove that when the signal is strong in the sense that the signal to noise ratio is above some threshold, our estimator is consistent and outperforms over many state-of-the-art algorithms. Moreover, our estimator is adaptive to the data set and does not require the variance of the noise to be known or estimated. Numerical simulations indicate that ASSVD still works well when the signal matrix is not very sparse.

• 한국멀티미디어학회논문지
• /
• 제22권1호
• /
• pp.35-43
• /
• 2019

Template tracking refers to the procedure of finding the most similar image patch corresponding to the given template through an image sequence. In order to obtain more accurate trajectory of the template, the template requires to be updated to reflect various appearance changes as it traverses through an image sequence. To do that, appearance images are used to model appearance variations and these are obtained by the computation of the principal components of the augmented image matrix at every iteration. Unfortunately, it is prohibitively expensive to compute the principal components at every iteration. Thus in this paper, we suggest a new Sliding Window based truncated URV Decomposition (TURVD) algorithm which enables updating their structure by recycling their previous decomposition instead of decomposing the image matrix from the beginning. Specifically, we show an efficient algorithm for updating and downdating the TURVD simultaneously, followed by the rank-one update to the TURVD while tracking the decomposition error accurately and adjusting the truncation level adaptively. Experiments show that the proposed algorithm produces no-meaningful differences but much faster execution speed compared to the typical algorithms in template tracking applications, thereby maintaining a good approximation for the principal components.

• 청정기술
• /
• 제22권4호
• /
• pp.299-307
• /
• 2016

These days, coal is one of the most important energy resources used for transportation, industry, and electricity. There are two types of coal: high-rank and low-rank. Low-rank coal has a low calorific value and contains large amounts of useless moisture. The quality of low-rank coal can be increased by simple drying technology and it needs to be stabilized by hydrocarbons (e.g. palm acid oil, PAO) to prevent spontaneous combustion and moisture re-adsorption. Spontaneous combustion becomes a major problem during coal mining, storage, and transportation. It can involve the loss of life, property, and economic value; reduce the quality of the coal; and increase greenhouse gas emissions. Besides spontaneous combustion, moisture re-adsorption also leads to a decrease in quality of the coal due to its lower heating value. In this work, PAO was used for additive to stabilize the upgraded coal. The objectives of the experiments were to determine the stabilization characteristic of coal by analyzing the behavior of upgraded coal by drying and PAO addition regarding crossing-point temperature of coal, the moisture behavior of briquette coal, and thermal decomposition behavior of coal.

• ETRI Journal
• /
• 제40권4호
• /
• pp.421-434
• /
• 2018

The application of deep neural networks (DNNs) to connect the world with cyber physical systems (CPSs) has attracted much attention. However, DNNs require a large amount of memory and computational cost, which hinders their use in the relatively low-end smart devices that are widely used in CPSs. In this paper, we aim to determine whether DNNs can be efficiently deployed and operated in low-end smart devices. To do this, we develop a method to reduce the memory requirement of DNNs and increase the inference speed, while maintaining the performance (for example, accuracy) close to the original level. The parameters of DNNs are decomposed using a hybrid of canonical polyadic-singular value decomposition, approximated using a tensor power method, and fine-tuned by performing iterative one-shot hybrid fine-tuning to recover from a decreased accuracy. In this study, we evaluate our method on frequently used networks. We also present results from extensive experiments on the effects of several fine-tuning methods, the importance of iterative fine-tuning, and decomposition techniques. We demonstrate the effectiveness of the proposed method by deploying compressed networks in smartphones.

• 대한수학회보
• /
• 제54권1호
• /
• pp.299-318
• /
• 2017

Sextic moment problems with an infinite algebraic variety are still widely open. We study the problem with a single cubic column relation associated to 3 parallel lines in which the variety is infinite. It turns out that this specific column relation has a strong connection with moment problems that have a symmetric algebraic variety. We present more concrete solutions to some sextic moment problems with a symmetric variety.