针对现有标签分布学习(Label Distribution Learning,LDL)算法较少考虑标签间关联性的问题,提出一种融合结构化标签依赖性的LDL算法.算法分为扩展、学习和恢复三个阶段:在扩展阶段,结合成对标签之间的关联性,构建结构化标签依赖性;在学...针对现有标签分布学习(Label Distribution Learning,LDL)算法较少考虑标签间关联性的问题,提出一种融合结构化标签依赖性的LDL算法.算法分为扩展、学习和恢复三个阶段:在扩展阶段,结合成对标签之间的关联性,构建结构化标签依赖性;在学习阶段,结合该依赖性,构建学习框架;在恢复阶段,利用最小二乘法求解超定方程组以预测标签分布.与七种常用的标签分布学习算法相比,在八个开放数据集上进行实验,提出的算法在Euclidean距离、Sørensen距离、Squardχ2距离、Kullback‐Leibler散度、Intersection相似度和Fidelity相似度六个主流评估指标上明显占优.展开更多
目的:构建携带活化T细胞表达和分泌调节因子(regulated upon activation normal T-cell expressed and secreted,RANTES/CCL5)基因及氧依赖性降解结构域(oxygen-dependent degradation domain,ODD)融合基因的重组腺病毒,并观察其体外趋...目的:构建携带活化T细胞表达和分泌调节因子(regulated upon activation normal T-cell expressed and secreted,RANTES/CCL5)基因及氧依赖性降解结构域(oxygen-dependent degradation domain,ODD)融合基因的重组腺病毒,并观察其体外趋化活性。方法:PCR法将人RANTES基因与ODD融合,构建携带该融合基因的重组腺病毒SG511-CCL5-ODD;增殖实验观察重组腺病毒增殖特性,ELISA法观察常氧和缺氧条件下RANTES蛋白的表达;趋化试验观察重组腺病毒感染肝癌细胞后的趋化活性。结果:成功构建携带人RANTES-ODD融合基因的重组腺病毒SG511-CCL5-ODD;增殖实验表明重组腺病毒具有肿瘤选择性复制的特性;缺氧条件下重组病毒转染肝癌细胞后RANTES蛋白表达量均比常氧条件下高(P<0.05),显示ODD可有效调节RANTES蛋白表达;趋化试验表明重组腺病毒感染肝癌细胞具有趋化NK92细胞的作用。结论:重组腺病毒SG511-CCL5-ODD体外能有效感染肝癌细胞株HepG2和Hep3B,并在ODD调控下表达RANTES蛋白,有效发挥体外趋化NK92细胞的活性。展开更多
Surface effects play an important role in the mechanical behavior of nanosized structural elements owing to the increased ratio of surface area to volume. The surface effects on the large deflection of nanowires were ...Surface effects play an important role in the mechanical behavior of nanosized structural elements owing to the increased ratio of surface area to volume. The surface effects on the large deflection of nanowires were considered. Both geometric nonlinearity in finite deformation and surface effects at nanoscale were taken into account to analyze the bending of nanowires subjected to a concentrated force. For simply supported beams and clamped-clamped beams, the influence of surface effects and geometric nonlinearity were discussed in detail. It is found that both surface effects and geometric nonlinearity tend to decrease the deflection of bending nanowires and thus increase the effective elastic modulus of nanowires. Surface effects yield the size dependent behavior of nanowires.展开更多
文摘针对现有标签分布学习(Label Distribution Learning,LDL)算法较少考虑标签间关联性的问题,提出一种融合结构化标签依赖性的LDL算法.算法分为扩展、学习和恢复三个阶段:在扩展阶段,结合成对标签之间的关联性,构建结构化标签依赖性;在学习阶段,结合该依赖性,构建学习框架;在恢复阶段,利用最小二乘法求解超定方程组以预测标签分布.与七种常用的标签分布学习算法相比,在八个开放数据集上进行实验,提出的算法在Euclidean距离、Sørensen距离、Squardχ2距离、Kullback‐Leibler散度、Intersection相似度和Fidelity相似度六个主流评估指标上明显占优.
文摘目的:构建携带活化T细胞表达和分泌调节因子(regulated upon activation normal T-cell expressed and secreted,RANTES/CCL5)基因及氧依赖性降解结构域(oxygen-dependent degradation domain,ODD)融合基因的重组腺病毒,并观察其体外趋化活性。方法:PCR法将人RANTES基因与ODD融合,构建携带该融合基因的重组腺病毒SG511-CCL5-ODD;增殖实验观察重组腺病毒增殖特性,ELISA法观察常氧和缺氧条件下RANTES蛋白的表达;趋化试验观察重组腺病毒感染肝癌细胞后的趋化活性。结果:成功构建携带人RANTES-ODD融合基因的重组腺病毒SG511-CCL5-ODD;增殖实验表明重组腺病毒具有肿瘤选择性复制的特性;缺氧条件下重组病毒转染肝癌细胞后RANTES蛋白表达量均比常氧条件下高(P<0.05),显示ODD可有效调节RANTES蛋白表达;趋化试验表明重组腺病毒感染肝癌细胞具有趋化NK92细胞的作用。结论:重组腺病毒SG511-CCL5-ODD体外能有效感染肝癌细胞株HepG2和Hep3B,并在ODD调控下表达RANTES蛋白,有效发挥体外趋化NK92细胞的活性。
基金Project(11072186)supported by the National Natural Science Foundation of China
文摘Surface effects play an important role in the mechanical behavior of nanosized structural elements owing to the increased ratio of surface area to volume. The surface effects on the large deflection of nanowires were considered. Both geometric nonlinearity in finite deformation and surface effects at nanoscale were taken into account to analyze the bending of nanowires subjected to a concentrated force. For simply supported beams and clamped-clamped beams, the influence of surface effects and geometric nonlinearity were discussed in detail. It is found that both surface effects and geometric nonlinearity tend to decrease the deflection of bending nanowires and thus increase the effective elastic modulus of nanowires. Surface effects yield the size dependent behavior of nanowires.
