We introduce novel profile-based string kernels for use with support vector machines (SVMs) the problems of protein classification and remote homology detection. These probabilistic profiles, such as those produced by PSI-BLAST algorithm, to define position-dependent mutation neighborhoods along sequences inexact matching k-length subsequences ("k-mers") in data. By an efficient data structure, are fast compute once profiles have been obtained. For example, time needed run order build is...

Inspired by their tremendous success in optical image detection and classification, convolutional neural networks (CNNs) have recently been used synthetic aperture radar automatic target recognition (SAR-ATR). Although CNN-based methods can achieve excellent performance, it is difficult to collect a large number of real SAR images available for training. In this paper, we introduce simulated data alleviate the problem insufficient training data. To address domain shift task transfer problems...

A variety of machine learning approaches have been applied to synthetic aperture radar (SAR) automatic target recognition. The performances these rely strongly on the quality and quantity training data. In real-world applications, however, it is challenging obtain sufficient data suitable for approaches. To alleviate this problem, a novel deep generative model SAR image generation proposed, which an extension Wasserstein autoencoder. network structure reconstruction loss function improved...

Abstract. Source-oriented chemical mechanisms enable direct source apportionment of air pollutants by explicitly representing precursor emissions and their reaction products in atmospheric models. These use source-tagged species to track evolution. However, scalability was previously limited the large number reactions required for interactions between two tagged species, such as NOx-NOx or VOC-NOx reactions. This study improves computational efficiency with a new method tracks total...

Abstract Based on two data mining techniques, this paper explores the integration path of college school management and ideological political education in aspects. First, ID3 decision tree algorithm is applied financial aid, by analyzing summarizing such as total amount students’ cafeteria consumption, number times, average consumption value, etc., used to get prediction model poor students. Secondly, plain Bayesian classification academic early warning, through analysis a series behavioral...

Numerous synthetic aperture radar-automatic target recognition (SAR-ATR) methods require a large amount of training data. However, collecting SAR data is both expensive and complicated in practical applications. Recognition with the limited has become vital issue SAR-ATR. To solve this problem, we propose model combining probabilistic inference meta-learning to transfer prior knowledge from simulated real First, use various tasks drawn learn global parameters model. Second, draw new...

Deep learning methods have achieved state-of-the-art performance on synthetic aperture radar (SAR) target recognition tasks in recent years. However, obtaining sufficient SAR images for training these deep is costly time and labor. This paper focuses recognizing targets with a few samples, that is, few-shot recognition. We combine neural networks' powerful feature representation capabilities the nonparametric flexibility of Gaussian processes (GPs) propose model based kernel learning....

To further enhance the early diagnose precision and efficiency for rotor crack under condition of strongly noise, AE signals is applied to extract fault feature cracks with advantages strong anti-noise ability, clear feature. Based on Pseudo Wigner-Ville Distribution (PWVD), amplitude frequency as vector are extracted from signal in this paper, which used predict types different depth by support machine (SVM). The experiment done verify proposed method diagnosis cracked rotor.

The fractured rock mass in the western cold region is affected by freezing and thawing disasters prone to local damage fracture along fissures’ ends. fatigue induced repeated frost heave traffic loads seriously endangers stability of roadbed. This paper selects sandstone as research object. Firstly, 20 freeze–thaw cycles were performed on samples with different inclination angles 30°, 45°, 60°, 90°. Subsequently, triaxial compression loading tests conducted explore mechanical properties...

A new method of displacement back analysis, named SVM-CTS, was proposed based on support vector machine (SVM) and continuous tabu search (CTS). On the one hand, SVM-CTS used SVM to build nonlinear mapping relationship between measuring point displacements rock soil mechanics parameters positive analysis study samples. other global optimization performance CTS catch optimal in space. The built by can fit forecast under different with high accuracy. prevent object function from trapping local...

This paper presents a new method of loess shoulder line extraction with openness and threshold segmentation.Openness is an angular measurement the relationship between surface relief horizontal distance which can be derived from DEM.As it has outstanding ability expressing positive negative terrain characteristics attribute ofanalysis-scale flexibility,in this study,we developed idea based on difference images threshould segmentation.Firstly,we calculated our study area DEM.Then,a image was...

本文分别采用斯托克斯沉降速率公式和重复深度吸管法计算了2005年4月、5月间在太湖进行的四次静沉降模拟实验中的沉降速度.结果表明:1)太湖水体中悬浮物的沉降属于絮凝沉降.2)水体中悬浮物浓度与沉降时间均呈现出明显的指数衰减规律(R²>0.80),悬浮物中无机物含量较高时这种规律更为明显(R²≥0.99).3)悬浮物浓较低时,太湖悬浮物的沉降速率与水体中的悬浮物浓度无明显的相关关系;而悬浮物浓度较高时,沉降速率随悬浮物浓度升高而增大.经拟合沉降速度(ω)与悬浮物浓度(C)之间符合Logistic曲线ω=0.021/(1+exp(-0.026(C-166.3))),R²=0.98,n= 54.4),斯托克斯公式可用来粗略估算太湖悬浮物的沉降速率,而重复深度吸管法则适合于较精确地计算太湖悬浮物的沉降速率.但在计算时须注意根据悬浮物的特性,选取其特征沉降速率.本文计算得到的太湖悬浮物的沉降速率范围为0.002 cm/s-0.005 cm/s.;Four experiments were conducted in...