Earticle

In this paper, fast image search with efficient additive kernels and kernel locality-sensitive hashing has been proposed. As to hold the kernel functions, recent work has probed methods to create locality-sensitive hashing , which guarantee our approach’s linear time, however existing methods still do not solve the problem of locality-sensitive hashing (LSH) and indirectly sacrifice the loss in accuracy of search results in order to allow fast queries. To improve the search accuracy, we show how to apply explicit feature maps into the homogeneous kernels, which help in feature transformation and combine it with kernel locality-sensitive hashing. We prove our method on several large datasets, and illustrate that it improve the accuracy relative to commonly used methods and make the task of object classification, content-based retrieval more fast and accurate.

According to the error approximation problem the sparse preserving projections (SPP) reconstruct the original sample. This paper proposes the algorithms based on orthogonal sparse preserving projections of kernel. In order to get sparse representation coefficients that contain more identification information by kernel method, it mapped samples to high-dimensional feature space to. Then, reconstructing sparse coefficient of kernel sparse representation increase the similar non neighbor sample weight, and reduce heterogeneous neighbor sample weight. Finally, the whole orthogonal constraint transformation improve the ability of sparse retain sample. The algorithm experiments were carried out on the YALE_B and ORL face database, and the recognition rate reached 96.3%, and the results verify the effectiveness and robustness of the algorithm.

Clustering is one of the main tasks used in pattern recognition and classification. Out of many methods that have been reported till date the most widely used methods are based on likelihood approach of mixture model. Among different mixture models, Expectation Maximization for Gaussian Mixture is most exploited and trusted algorithm for data clustering. However, it has some short comings such as initial parameters are to be given a-priori, convergence speed is slow and the results obtained are highly dependent upon the initial parameters. Many variations have been carried out in implementing EM algorithm but still there is ample scope for improvement. The proposed algorithm tries to overcome these shortcomings and provide more robust and efficient version of clustering algorithm. An improvement related to cluster partitioning is proposed in the existing algorithm resulting some advantages. The robustness and efficacy of the algorithm is demonstrated qualitatively as well as quantitatively with the help of some experiments.

This paper addresses a problem of classifying human breast cancer into its subtypes. A main ingredient in our approach is kernel machines such as support vector machine (SVM). kernel principal component analysis (KPCA). and kernel partial least squares (KPLS). In the task of breast cancer classification, we employ both SVM and KPLS and compare their results. In addition to this classification. we also analyze the patterns of clinical outcomes in the feature space. In order to visualize the clinical outcomes in low-dimensional space, both KPCA and KPLS are used. It turns out that these methods are useful to identify correlations between clinical outcomes and the nonlinearly protected expression profiles in low-dimensional feature space.

The purpose of the ensemble methods is to increase the accuracy of prediction through combining many classifiers. According to recent studies, it is proved that random forests and forward stagewise regression have good accuracies in classification problems. However they have great prediction error in separation boundary points because they used decision tree as a base learner. In this study, we use the kernel ridge regression instead of the decision trees in random forests and boosting. The usefulness of our proposed ensemble methods was shown by the simulation results of the prostate cancer and the Boston housing data.

Background and Objectives: Increase of mortality rates of gastric cancer in Iran and the world in recent years reveal necessity of studies on this disease. Here, hazard function for gastric cancer patients was estimated using Wavelet and Kernel methods and some related factors were assessed. Materials and Methods: Ninety-five gastric cancer patients in Fayazbakhsh Hospital between 1996 and 2003 were studied. The effects of age of patients, gender, stage of disease and treatment method on patient's lifetime were assessed. For data analyses, survival analyses using Wavelet method and Log-rank test in R software were used. Results: Nearly 25.3% of patients were female. Fourteen percent had surgery treatment and the rest had treatment without surgery. Three fourths died and the rest were censored. Almost 9.5% of patients were in early stages of the disease, 53.7% in locally advance stage and 36.8% in metastatic stage. Hazard function estimation with the wavelet method showed significant difference for stages of disease (P<0.001) and did not reveal any significant difference for age, gender and treatment method. Conclusion: Only stage of disease had effects on hazard and most patients were diagnosed in late stages of disease, which is possibly one of the most reasons for high hazard rate and low survival. Therefore, it seems to be necessary a public education about symptoms of disease by media and regular tests and screening for early diagnosis.

We propose and analyze spectral and pseudo-spectral Jacobi-Galerkin approaches for weakly singular Volterra integral equations (VIEs). We provide a rigorous error analysis for spectral and pseudo-spectral Jacobi-Galerkin methods, which show that the errors of the approximate solution decay exponentially in $L^{\infty}$ norm and weighted $L^2$-norm. The numerical examples are given to illustrate the theoretical results.

In this article, a homotopy perturbation transform method (HPTM) and the Laplace transform combined with Taylor expansion method are presented for solving Volterra integral equations with a convolution kernel. The (HPTM) is innovative in Laplace transform algorithm and makes the calculation much simpler while in the Laplace transform and Taylor expansion method we first convert the integral equation to an algebraic equation using Laplace transform then we find its numerical inversion by power series. The numerical solution obtained by the proposed methods indicate that the approaches are easy computationally and its implementation very attractive. The methods are described and numerical examples are given to illustrate its accuracy and stability.

최근 들어 기계학습 분야의 여러 우수 방법론들은 관련 분야에서 우수한 성능을 보이며 계속적으로 빠르게 발전하고 있다. 커널 기법은 오랫동안 이러한 경쟁에 동참하며 우수한 성능을 산출함으로써 많은 주목을 끌어왔던 방법론 중 하나이다. 그 중에서도 확률론적 커널 기법인 가우시안 프로세스는 함수 근사, 패턴 분류, 비선형 특징 검출 등의 기계학습 문제에서 뛰어난 예측 및 오차 범위 결과를 제공하고 있다. 본 논문에서는 보행 중 낙상이 발생하는 경우를 탐지하기 위하여 웨어러블 센서와 가우시안 프로세스 및 커널 탐색 등을 이용하는 방안을 고려하였다. 이러한 연구에 있어서, 가우시안 프로세스 모델링 단계에서는, 가우시안 프로세스 기반 동적 모델링과 자동 커널 탐색을 고려하였다. 가우시안 프로세스 기반 동적 모델링에서는 인덱스의 역할을 상태 벡터가 수행하는 반면에, 자동 커널 탐색에서는 시간이 이러한 역할을 수행한다. 고려된 방법론의 응용 가능성을 관찰하기 위하여 실험을 실시하였고, 가능성 있는 결과를 관찰하였다.

Recently, many advanced machine learning methods have demonstrated excellent performance in many related fields. The kernel methods have been one of such advanced methodologies that have attracted a great deal of attention because they have long been involved in this competition and producing superior performance. Among them, the Gaussian process, which is a probabilistic kernel method particularly effective for machine learning problems such as function approximation, pattern classification, and non-linear feature extraction, is providing excellent prediction and error-bar results. In this paper, we consider the problem of detecting the occurrence of fall during walking based on wearable sensors, Gaussian process methodologies, and kernel search. Here in the Gaussian process modeling stage, Gaussian process-based dynamic modeling and automatic kernel search were utilized. In Gaussian process-based dynamic modeling, the latent state vector performs the role of the index, while in the automatic kernel search, time plays the index role. Experiments were conducted to observe the applicability of the proposed methodology, and promising outcomes were observed.

This study tries to know the extent of intra-industry trade(IIT), vertical intra-trade industry(VIIT), and horizontal intra-trade industry(HIIT) in Korea's foreign trade with Russia and studies the deterministic factors to these. In the analysis of various deterministic factors to these, the existent researches mainly used OLS or GLS method which has many innate problems by assuming many conditions in advance. This study to overcome these problems uses GLS and Kernel method complementarily and derives the conclusions. The main findings of this study are as follow: The trade pattern between Korea and Russia is mainly vertical intra-industry trade(VIIT) but the influence of country-specific deterministic factors goes to horizontal intra-industry trade(HIIT) very much.

본 연구는 식용옥수수 풋이삭의 식감의 결정에 영향을 미치는 과피의 두께를 측정하는 방법을 규명하기 위하여 과피 시료의 준비 방법에 따른 과피 두께의 차이와 등숙 시기별 과피 두께의 변화를 조사하기 위하여 수행하였다. 1. 찰옥수수와 초당옥수수 각 3품종의 시료를 수확 직후 생옥수수(생시료) 알곡을 바로 측정하거나 상온의 음지에서 완전(건조시료) 건조시키거나 $-4^{\circ}C$로 냉동시킨 후 과피를 측정하였다. 2. 시료 준비 방법과 등숙 시기에 따른 삼원 교호작용의 유의성이 인정되어 어느 한 요인에 대한 일반적인 결론을 도출하기 보다는 각 시료 준비 방법과 특정 등숙 시기에 따른 과피 두께의 비교를 수행하였다. 3. 초당옥수수는 건조시료에서 등숙이 진행됨에 따라 비교적 안정된 과피 두께 측정값을 얻을 수 있었으며 찰옥수수는 비해 모든 시료 준비 방법에서 초당옥수수보다 안정된 값을 보였다. 특히 풋찰옥수수 수확 적기인 수정 후 24일경에는 시료 준비 방법에 따른 차이는 크게 나타나지 않을 것으로 기대된다.

Pericarp thickness of vegetable corns such as sweet and waxy corn is one of the crucial traits, contributing to their edible quality. This study was carried out to compare the pericarp thickness of super sweet and waxy corn hybrids measured with kernel samples prepared using different methods at different grain filling periods. The samples comprised excised pericarp from dried, frozen (at $-4^{\circ}C$), and fresh kernels. Analysis of variance performed separately on super sweet and waxy corn hybrids indicated a significant three-way interaction among cultivars, kernel sample preparation methods, and days after pollination (DAP). Dried samples of super sweet corn hybrids presented reasonably stable pericarp thickness measurements during grain filling, while all the sample preparation methods fluctuated less as grains of waxy corn hybrids matured. Waxy corn is best consumed at around 24 days after pollination. Pericarp thickness of waxy kernel samples regardless of preparation methods investigated was the same at 24 DAP with a few exceptions. Overall, the common method of drying kernel samples before pericarp excision can provide reliable data for estimating the tenderness of vegetable corn hybrids.