利用带无标签数据的双支持向量机对恒星光谱分类

恒星光谱分类是天文技术与方法领域一直关注的热点问题之一。随着观测设备持续运行和不断改进,人类获得的光谱数量与日俱增。这些海量光谱为人工处理带来了极大挑战。鉴于此,研究人员开始关注数据挖掘算法,并尝试对这些光谱进行数据挖掘。近年来,神经网络、自组织映射、关联规则等数据挖掘方法广泛应用于恒星光谱分类。在这些方法中,支持向量机(SVM)以其强大的学习能力和高效的分类性能而备受推崇。SVM的基本思想是试图在两类样本之间找到一个最优分类面将两类分开。SVM在求解时,通过将其最优化问题转化为具有(QP)形式的凸问题,进而得到全局最优解。尽管该方法在实际应用中表现优良,但为了进一步提高其分类能力,有的学者提出双支持向量机(TSVM)。该方法通过构造两个非平行的分类面将两类分开,每一类靠近某个分类面,而远离另一个分类面。TSVM的计算效率较之传统SVM提高近4倍,因此,自TSVM提出后便受到研究人员的持续关注,并出现若干改进算法。在恒星光谱分类中,一般分类算法都是根据历史观测光谱来建立分类模型,其中最关键的是对光谱进行人工标注,这项工作极为繁琐,且容易犯错。如何利用已标记的光谱以及部分无标签的光谱来建立分类模型显得尤为重要。因此,提出带无标签数据的双支持向量机(TSVMUD)用以实现对恒星光谱智能分类的目的。该方法首先将光谱分为训练数据集和测试数据集两部分;然后,在训练集上进行学习,得到分类依据;最后利用分类依据对测试集上的光谱进行验证。继承了双支持向量机的优势,更重要的是,在训练集上学习分类模型过程中,不仅考虑有标记的训练样本,也考虑部分未标记的样本。一方面提高了学习效率,另一方面得到更优的分类模型。在SDSS DR8恒星光谱数据集上的比较实验表明,与支持向量机SVM、双支持向量机TSVM以及K近邻(KNN)等传统分类方法相比,带无标签数据的双支持向量机TSVMUD具有更优的分类能力。然而,该方法亦存在一定的局限性,其中一大难题是其无法处理海量光谱数据。该工作将借鉴海量数据随机采样思想,利用大数据处理技术,来对所提方法在大数据环境下的适应性展开进一步研究。     

双扫描激光雷达精细探测低层大气气溶胶方法

为实现水平非均匀分布的低层大气气溶胶光学特性的遥感探测,提出一种基于双扫描激光雷达的气溶胶精细探测方法.该方法以扫描激光雷达为遥感探测工具,通过双激光雷达相向交叉扫描工作模式,实现对同一空域近地表气溶胶全视野剖面的交叉探测,从而提供双激光雷达方程组以精确求解气溶胶消光和后向散射系数.在数据反演过程中,通过对交叉扫描区域进行坐标化和网格化处理、网格像素单元的初值预设,以及双扫描激光雷达方程组的数值逼近反演得到气溶胶消光.利用长距离扫描激光雷达的数据对该方法进行验证,结果表明,该方法与多角度方法反演所得到的结果随高度变化的趋势具有高度的一致性;同时双扫描激光雷达可提供交叉扫描区域剖面的气溶胶浓度分布,相比于单条廓线具有较大的优势.     

Quantitative Surface Chirality Detection with Sum Frequency Generation Vibrational Spectroscopy: Twin Polarization Angle Approach

介绍了一种新的表面和频振动光谱方法: 定量手性检测中的双偏振角方法. 一般来讲表面光谱信号中非手性项贡献最强,而纯界面手性信号要比非手性信号弱两到三个数量级. 因此在和频振动光定量测量和分析界面手性的贡献问题上存在困难. 在和频振动光谱的双偏振角方法中,通过固定入射红外光偏振方向,并同时改变入射可见光和初涉和频信号光的偏振角进行测量. 这样所得到的偏振依赖的和频振动光谱信号能够直接给出界面手性贡献的特征,并 且能够用于准确测量界面的手性和非手性贡献. 对双偏振角方法进行了描述,并且通过S和R手性的柠檬烯空     

On the Ingredients of the Twin Paradox

If M is a 4-dimensional connected, orientable flat spacetime manifold endowed with a time-arrow, and if the existence of a finite speed limit to energy/information transfers over M can be assessed, then the “twin paradox” necessarily follows (and indeed, the full special relativity theory). Two other implicit ingredients of the paradox are also identified.     

双螺杆冻胶纺丝法制备有机液体吸附功能纤维及其性能研究

在聚合釜中合成了BMA/HEMA二元共聚树脂,利用双螺杆纺丝机、采用冻胶纺丝技术制备了有机液体吸附功能纤维,研究了单体投料比中HEMA质量分数对纤维各种溶胀行为的影响,利用动态热机械分析仪(DMA)对纤维的动态力学性能进行了研究,同时利用环境扫描电子显微镜(SEM)观察了纤维的表面和断面形貌.结果表明,HEMA引入大分子后,所得纤维对有机液体甲苯、三氯乙烯具有吸附功能,且吸附过程符合sigmoidal模型中的Hill方程;随HEMA质量分数的增加,纤维对甲苯和三氯乙烯的饱和吸附量增大,对甲苯和三氯乙烯的最大吸附量分别可达10g/g和21g/g;随HEMA质量分数的增加,纤维剩余率增加的同时,使纤维对甲苯和三氯乙烯的握持能力增强;HEMA质量分数对纤维动态力学性能有突出影响,特别是链段运动受其影响更为明显;HEMA质量分数对纤维表面和断面形貌均有显著影响,特别是当HEMA质量分数为15wt%时,所得纤维表面出现了许多空洞,其断面存在许多尺寸不均的空洞.     

Synthesis and Pregnancy Terminating Activity of 2-Aryl imidazo [2,1-a] isoquinolines

Several kinds of potential pregnancy terminating agents have been identified such as 2-aryl pyrazolo [5,1-a] isoindoles and isoquinolines1, pyrazolo[1,5-a] indoles and quinolines2, 3,5-diaryl-s-triazoles3, 2-phenyl triazole, pyrozole, imidazole isoindoles and homologues4 etc. In order to investigate their structure-activity relationship, the quantitative structure-activity relationship (QSAR) of 2-aryl imidazo [2,1-a] isoquinolines5 has been studied with multiple regression and BP Artificial…     

Effect of Sex Steroids and PGF2α on the Expression of Their Receptors and Decorin in Bovine Caruncular Epithelial Cells in Early–Mid Pregnancy

Changes in the expression of various genes, including pregnancy-associated hormone receptors and extracellular matrix proteins, have been suggested to play a significant role in bovine placental development. This study aimed to examine the influence of sex steroids and PGF_2α on decorin (DCN) expression in the epithelial cells of bovine caruncle in early–mid pregnancy in cows. The expression patterns of DCN, PTGFR_, PGR and ESR1 were analyzed by RT-qPCR and Western blotting in primary caruncular epithelial cell cultures (PCECC) and placental tissue homogenates derived from the 2nd and 4th months of pregnancy. PCECC were found to express DCN, PTGFR_, PGR and ESR1. The intensity of PGR staining was higher in cells derived from the 4th month of pregnancy (p < 0.05). The 17β-estradiol, progesterone and PGF_2α have not been shown to affect DCN expression. PGF_2α decreased PTGFR expression in cells derived from the 4th month of gestation (p < 0.05). In conclusion, the results of the present preliminary study showed that the expression of the PTGFR, ESR1, PGR and DCN in PCECC does not vary throughout early–mid pregnancy. Further studies should be carried out to observe the relationship between hormonal status and cellular adhesion to determine their importance for properly developing placentation and pregnancy in cows.     

Maternal Physiological Variations Induced by Chronic Gestational Hypoxia: 1H NMR-Based Metabolomics Study

Metabolomics have been widely used in pregnancy-related diseases. However, physiological variations induced by chronic hypoxia during pregnancy are not well characterized. We aimed to investigate physiological variations induced by chronic hypoxia during pregnancy. A Sprague–Dawley (SD) pregnant rat model of chronic hypoxia was established. Plasma and urine metabolite profiles at different stages of the pregnancy were detected by ¹H NMR (nuclear magnetic resonance). Multivariate statistical analysis was used to analyze changes in plasma and urine metabolic trajectories at different time-points. We identified hypoxia-induced changes in the levels of 30 metabolites in plasma and 29 metabolites in urine during different stages of pregnancy; the prominently affected metabolites included acetic acid, acetone, choline, citric acid, glutamine, isoleucine, lysine, and serine. Most significant hypoxia-induced changes in plasma and urine sample metabolites were observed on the 11th day of gestation. In summary, chronic hypoxia has a significant effect on pregnant rats, and may cause metabolic disorders involving glucose, lipids, amino acids, and tricarboxylic acid cycle. Metabolomics study of the effect of hypoxia during pregnancy may provide insights into the pathogenesis of obstetric disorders.