• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권5호
• /
• pp.2082-2102
• /
• 2018

In recent years, social network related applications such as WeChat, Facebook, Twitter and so on, have attracted hundreds of millions of people to share their experience, plan or organize, and attend social events with friends. In these operations, plenty of valuable information is accumulated, which makes an innovative approach to explore users' preference and overcome challenges in traditional recommender systems. Based on the study of the existing social network recommendation methods, we find there is an abundant information that can be incorporated into probability matrix factorization (PMF) model to handle challenges such as data sparsity in many recommender systems. Therefore, the research put forward a unified social network recommendation framework that combine tags, trust between users, ratings with PMF. The uniformed method is based on three existing recommendation models (SoRecUser, SoRecItem and SoRec), and the complexity analysis indicates that our approach has good effectiveness and can be applied to large-scale datasets. Furthermore, experimental results on publicly available Last.fm dataset show that our method outperforms the existing state-of-art social network recommendation approaches, measured by MAE and MRSE in different data sparse conditions.

• Journal of Ubiquitous Convergence Technology
• /
• 제2권1호
• /
• pp.18-26
• /
• 2008

The invariance relation existing in the non-negative matrix factorization (NMF) is used for constructing robust image hashes in this work. The image is first re-scaled to a fixed size. Low-pass filtering is performed on the luminance component of the re-sized image to produce a normalized matrix. Entries in the normalized matrix are pseudo-randomly re-arranged under the control of a secret key to generate a secondary image. Non-negative matrix factorization is then performed on the secondary image. As the relation between most pairs of adjacent entries in the NMF's coefficient matrix is basically invariant to ordinary image processing, a coarse quantization scheme is devised to compress the extracted features contained in the coefficient matrix. The obtained binary elements are used to form the image hash after being scrambled based on another key. Similarity between hashes is measured by the Hamming distance. Experimental results show that the proposed scheme is robust against perceptually acceptable modifications to the image such as Gaussian filtering, moderate noise contamination, JPEG compression, re-scaling, and watermark embedding. Hashes of different images have very low collision probability. Tampering to local image areas can be detected by comparing the Hamming distance with a predetermined threshold, indicating the usefulness of the technique in digital forensics.

• 한국음향학회지
• /
• 제41권4호
• /
• pp.468-479
• /
• 2022

수중에서 능동소나를 이용하여 표적을 탐지하기 위하여, 송신음이 표적에 반사된 반향을 수신함으로써 표적의 위치를 감지한다. 이때 산란체로부터의 잔향이 발생하며, 이는 표적 반향의 탐지를 방해하게 된다. 효과적인 표적 탐지를 위해 자기회귀 모델기반의 백색화 기법이나 주성분역산 등의 잔향 제거 기법이 연구된 바 있으며, 최근에는 비음수 행렬 분해 기반의 기법이 고안되었다. 비음수 행렬 분해 기반의 잔향 제거 기법은 기존의 기법에 비해 향상된 성능을 보여주지만, 송수신기의 위치 및 거리에 의한 감쇠 등이 고려되지 않았다. 본 논문에서는, 양상태 소나에서 지속파 송신 파형을 사용하는 경우에 대하여 수신기의 방향성과 그에 관련된 도플러, 그리고 거리에 대한 감쇠 등의 전처리를 통해 성능을 개선하였다. 본 연구에서 고안된 시스템의 성능을 확인하기 위하여 잔향 모델을 이용한 시뮬레이션을 수행하였다, 시뮬레이션 결과 1 %의 낮은 오탐지율에서 기존의 비음수 행렬 분해 기법 대비 10 % ~ 40 %의 탐지율 성능 향상이 있음을 확인하였다.

• 한국산학기술학회논문지
• /
• 제12권11호
• /
• pp.5150-5155
• /
• 2011

인지증을 유발하는 원인은 알츠하이머병(Alzheimer's Disease: AD)과 혈관성 인지증(vascular Dementia: VD)이 가장 높은 비율을 차지한다. 본 논문에서는 측정된 라만 스펙트럼에서 AD, VD, 정상(NOR: normal)을 분류하기 위해 변별력 있는 영역을 조사하고, 특징 변환을 이용한 분류 실험 결과를 제시하였다. 혈소판으로부터 측정한 라만 스펙트럼은 먼저 smoothing을 적용한 다음 배경 잡음을 제거하고 스펙트럼의 기준 피크를 중심으로 그 위치를 정렬하였고 minmax 방법을 사용하여 정규화 하였다. 전처리를 거친 스펙트럼은 AD와 VD, NOR를 변별하기 가장 용이한 영역을 결정하기 위해 조사되었으며, 그 결과 725-777, 1504-1592, 1632-1700 $cm^{-1}$ 영역에서 스펙트럼이 많은 차이를 보임을 확인하였다. 분류 실험은 선택한 각 영역에 대하여 PCA(principal component analysis)와 NMF(nonnegative matrix factorization) 방법을 적용하여 얻은 특징을 이용하여 행하였다. 총 327개의 라만 스펙트럼에 대한 MAP(maximum a posteriori probability) 분류 실험 결과에 따르면, 본 연구에서 제안된 국부 영역 변환 특징을 사용했을 때 평균 92.8 %의 분류율을 보임을 알 수 있었다.

• ETRI Journal
• /
• 제38권3호
• /
• pp.487-493
• /
• 2016

Two new methods are proposed for an unsupervised adaptation of a language model (LM) with a single sentence for automatic transcription tasks. At the training phase, training documents are clustered by a method known as Latent Dirichlet allocation (LDA), and then a domain-specific LM is trained for each cluster. At the test phase, an adapted LM is presented as a linear mixture of the now trained domain-specific LMs. Unlike previous adaptation methods, the proposed methods fully utilize a trained LDA model for the estimation of weight values, which are then to be assigned to the now trained domain-specific LMs; therefore, the clustering and weight-estimation algorithms of the trained LDA model are reliable. For the continuous speech recognition benchmark tests, the proposed methods outperform other unsupervised LM adaptation methods based on latent semantic analysis, non-negative matrix factorization, and LDA with n-gram counting.

• Asian Journal of Atmospheric Environment
• /
• 제8권3호
• /
• pp.126-139
• /
• 2014

In order to estimate the influence of sources on $PM_{2.5}$ in the industrial area of Japan, we carried out a source analysis using chemical component data of $PM_{2.5}$. $PM_{2.5}$ samples were collected intermittently at an industrial area in Japan from July 2010 to November 2012. Water soluble ions ($Cl^-$, $NO_3{^-}$, $SO{_4}^{2-}$, $Na^+$,$NH_4{^+}$, $K^+$, $Mg^{2+}$, $Ca^{2+}$), elements (Al, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Cd, Sb, Pb), and carbonaceous species (OC, EC) of the $PM_{2.5}$ (a total of 198 samples) were analyzed. Positive Matrix Factorization (PMF) model was applied to the data of those chemical components to identify the source of $PM_{2.5}$. At this observation site, nine factors were extracted. The major contributors of $PM_{2.5}$ were secondary sulfate 1, in which loading factors of $SO{_4}^{2-}$ and $NH_4{^+}$ were large (percentage source contribution: 20.9%), traffic, in which loading factors of OC (organic carbon) and EC (elemental carbon) were large (20.8%), secondary sulfate 2, in which loading factors of K and $SO{_4}^{2-}$ were large (8.0%), steel mills (7.8%), secondary chloride and nitrate (7.0%), soil (5.0%), heavy oil combustion (3.8%), sea salt (3.8%), and coal combustion (2.3%). The conditional probability function (CPF) and the potential source contribution function (PSCF) were carried out to examine the influence of a regional source and a broad-based source, respectively. CPF results supported local source influences such as steel mills, sea salt, traffic, coal combustion, and heavy oil combustion. PSCF results suggested that ships in the East China Sea, an industrial area of the east coastal region of China, and an active volcano in the Kyushu region of Japan were potential regional sources of secondary sulfate 1. Secondary sulfate 2 was affected by the burning of biomass fields and by coal combustion in Chinese urban areas such as Beijing, Hebei, and western Inner Mongolia. Source characterization using continuous data from one site showed a potential source representing fossil fuel combustion is affected both by regional and broad-based sources.