Various networks exist in the world today including biological, social, information, and communication networks with the Internet as the largest network of all. One salient structural feature of these networks is the ...Various networks exist in the world today including biological, social, information, and communication networks with the Internet as the largest network of all. One salient structural feature of these networks is the formation of groups or communities of vertices that tend to be more connected to each other within the same group than to those outside. Therefore, the detection of these communities is a topic of great interest and importance in many applications and different algorithms including label propagation have been developed for such purpose. Speaker-listener label propagation algorithm (SLPA) enjoys almost linear time complexity, so desirable in dealing with large networks. As an extension of SLPA, this study presented a novel weighted label propagation algorithm (WLPA), which was tested on four real world social networks with known community structures including the famous Zachary's karate club network. Wilcoxon tests on the communities found in the karate club network by WLPA demonstrated an improved statistical significance over SLPA. Withthehelp of Wilcoxon tests again, we were able to determine the best possible formation of two communities in this network relative to the ground truth partition, which could be used as a new benchmark for assessing community detection algorithms. Finally WLPA predicted better communities than SLPA in two of the three additional real social networks, when compared to the ground truth.展开更多
在对科技领域视频文本进行分类时,容易忽略分类贡献度较高的专业名词。为此,改进传统Labeled潜在Dirichlet分布(LDA)模型,建立用于科技领域视频文本的M ul CHI-Labeled LDA模型,避免偏向高频词的现象。通过构建领域术语库以突出专业名词...在对科技领域视频文本进行分类时,容易忽略分类贡献度较高的专业名词。为此,改进传统Labeled潜在Dirichlet分布(LDA)模型,建立用于科技领域视频文本的M ul CHI-Labeled LDA模型,避免偏向高频词的现象。通过构建领域术语库以突出专业名词,同时使用卡方加权和文本位置加权算法提升主题词质量。实验结果表明,与Labeled LDA模型相比,该模型可以解决专业名词被忽略的问题,并能有效提高主题词质量和分类准确率。展开更多
针对现有的特征选择模型未涉及特征和标记集之间的相关度,造成分类精度偏低等情况,提出了基于ReliefF和最大相关最小冗余(maximum Relevance and Minimum Redundancy,mRMR)的多标记特征选择.首先,运用互信息计算每个标记和标记集之间的...针对现有的特征选择模型未涉及特征和标记集之间的相关度,造成分类精度偏低等情况,提出了基于ReliefF和最大相关最小冗余(maximum Relevance and Minimum Redundancy,mRMR)的多标记特征选择.首先,运用互信息计算每个标记和标记集之间的相关度,使用每项相关度占其相关度之和的比例设计了标记权重,由此构建了特征和标记集间的相关度,初选与标记集相关度高的特征;其次,计算对象在特征上的距离,构建了新的特征权值更新公式,基于标记权重改进多标记ReliefF模型.然后,基于互信息和标记权重构建了最大相关性,设计了最小冗余性及其新的最大相关最小冗余评价准则,并将其应用于多标记特征选择,进一步剔除冗余特征;最后,设计了一种基于ReliefF和最大相关最小冗余的多标记特征选择算法,有效提高了多标记分类性能.在8个多标记数据集上测试所提算法的平均分类精度、覆盖率、汉明损失、1错误率和排序损失,实验结果证明了该算法的有效性.展开更多
AIM: To assess intravoxel incoherent motion diffusionweighted imaging(IVIM-DWI) for monitoring early efficacy of chemotherapy in a human gastric cancer mouse model.METHODS: IVIM-DWI was performed with 12 b-values(0-80...AIM: To assess intravoxel incoherent motion diffusionweighted imaging(IVIM-DWI) for monitoring early efficacy of chemotherapy in a human gastric cancer mouse model.METHODS: IVIM-DWI was performed with 12 b-values(0-800 s/mm2) in 25 human gastric cancer-bearing nude mice at baseline(day 0), and then they were randomly divided into control and 1-, 3-, 5- and 7-d treatment groups(n = 5 per group). The control group underwent longitudinal MRI scans at days 1, 3, 5 and 7, and the treatment groups underwent subsequent MRI scans after a specified 5-fluorouracil/calciumfolinate treatment. Together with tumor volumes(TV), the apparent diffusion coefficient(ADC) and IVIM parameters [true water molecular diffusion coefficient(D), perfusion fraction(f) and pseudo-related diffusion coefficient(D*)] were measured. The differences in those parameters from baseline to each measurement(ΔTV%, ΔADC%, ΔD%, Δf% and ΔD*%) were calculated. After image acquisition, tumor necrosis, microvessel density(MVD) and cellular apoptosis were evaluated by hematoxylin-eosin(HE), CD31 and terminal-deoxynucleotidyl transferase mediated nick end labeling(TUNEL) staining respectively, to confirm the imaging findings. Mann-Whitney test and Spearman's correlation coefficient analysis were performed.RESULTS: The observed relative volume increase(ΔTV%) in the treatment group were significantly smaller than those in the control group at day 5(ΔTV_(treatment)% = 19.63% ± 3.01% and ΔTVcontrol% = 83.60% ± 14.87%, P = 0.008) and day 7(ΔTV_(treatment)% = 29.07% ± 10.01% and ΔTV_(control)% = 177.06% ± 63.00%, P = 0.008). The difference in ΔTV% between the treatment and the control groups was not significant at days 1 and 3 after a short duration of treatment. Increases in ADC in the treatment group(ΔADC%_(treatment), median, 30.10% ± 18.32%, 36.11% ± 21.82%, 45.22% ± 24.36%) were significantly higher compared with the control group(ΔADC%_(control), median, 4.98% ± 3.39%, 6.26% ± 3.08%, 9.24% ± 6.33%) at days 3, 5 and 7(P = 0.008, P = 0.016, P = 0.008, respectively). Increases in D in the treatment group(ΔD%_(treatment), median 17.12% ± 8.20%, 24.16% ± 16.87%, 38.54% ± 19.36%) were higher than those in the control group(ΔD%_(control), median-0.13% ± 4.23%, 5.89% ± 4.56%, 5.54% ± 4.44%) at days 1, 3, and 5(P = 0.032, P = 0.008, P = 0.016, respectively). Relative changes in f were significantly lower in the treatment group compared with the control group at days 1, 3, 5 and 7 follow-up(median,-34.13% ± 16.61% vs 1.68% ± 3.40%, P = 0.016;-50.64% ± 6.82% vs 3.01% ± 6.50%, P = 0.008;-49.93% ± 6.05% vs 0.97% ± 4.38%, P = 0.008, and-46.22% ± 7.75% vs 8.14% ± 6.75%, P = 0.008, respectively). D* in the treatment group decreased significantly compared to those in the control group at all time points(median,-32.10% ± 12.22% vs 1.85% ± 5.54%, P = 0.008;-44.14% ± 14.83% vs 2.29% ± 10.38%, P = 0.008;-59.06% ± 19.10% vs 3.86% ± 5.10%, P = 0.008 and-47.20% ± 20.48% vs 7.13% ± 9.88%, P = 0.016, respectively). Furthermore, histopathologic findings showed positive correlations with ADC and D and tumor necrosis(r_s = 0.720, P < 0.001; r_s = 0.522, P = 0.007, respectively). The cellular apoptosis of the tumor also showed positive correlations with ADC and D(r_s = 0.626, P = 0.001; r_s = 0.542, P = 0.005, respectively). Perfusionrelated parameters(f and D*) were positively correlated to MVD(r_s = 0.618, P = 0.001; r_s = 0.538, P = 0.006, respectively), and negatively correlated to cellular apoptosis of the tumor(r_s =-0.550, P = 0.004; r_s =-0.692, P < 0.001, respectively).CONCLUSION: IVIM-DWI is potentially useful for predicting the early efficacy of chemotherapy in a human gastric cancer mouse model.展开更多
文摘Various networks exist in the world today including biological, social, information, and communication networks with the Internet as the largest network of all. One salient structural feature of these networks is the formation of groups or communities of vertices that tend to be more connected to each other within the same group than to those outside. Therefore, the detection of these communities is a topic of great interest and importance in many applications and different algorithms including label propagation have been developed for such purpose. Speaker-listener label propagation algorithm (SLPA) enjoys almost linear time complexity, so desirable in dealing with large networks. As an extension of SLPA, this study presented a novel weighted label propagation algorithm (WLPA), which was tested on four real world social networks with known community structures including the famous Zachary's karate club network. Wilcoxon tests on the communities found in the karate club network by WLPA demonstrated an improved statistical significance over SLPA. Withthehelp of Wilcoxon tests again, we were able to determine the best possible formation of two communities in this network relative to the ground truth partition, which could be used as a new benchmark for assessing community detection algorithms. Finally WLPA predicted better communities than SLPA in two of the three additional real social networks, when compared to the ground truth.
文摘在对科技领域视频文本进行分类时,容易忽略分类贡献度较高的专业名词。为此,改进传统Labeled潜在Dirichlet分布(LDA)模型,建立用于科技领域视频文本的M ul CHI-Labeled LDA模型,避免偏向高频词的现象。通过构建领域术语库以突出专业名词,同时使用卡方加权和文本位置加权算法提升主题词质量。实验结果表明,与Labeled LDA模型相比,该模型可以解决专业名词被忽略的问题,并能有效提高主题词质量和分类准确率。
文摘针对现有的特征选择模型未涉及特征和标记集之间的相关度,造成分类精度偏低等情况,提出了基于ReliefF和最大相关最小冗余(maximum Relevance and Minimum Redundancy,mRMR)的多标记特征选择.首先,运用互信息计算每个标记和标记集之间的相关度,使用每项相关度占其相关度之和的比例设计了标记权重,由此构建了特征和标记集间的相关度,初选与标记集相关度高的特征;其次,计算对象在特征上的距离,构建了新的特征权值更新公式,基于标记权重改进多标记ReliefF模型.然后,基于互信息和标记权重构建了最大相关性,设计了最小冗余性及其新的最大相关最小冗余评价准则,并将其应用于多标记特征选择,进一步剔除冗余特征;最后,设计了一种基于ReliefF和最大相关最小冗余的多标记特征选择算法,有效提高了多标记分类性能.在8个多标记数据集上测试所提算法的平均分类精度、覆盖率、汉明损失、1错误率和排序损失,实验结果证明了该算法的有效性.
基金Supported by National Research Foundation of South Korea,No.NRF-2013R1A1A2013878 and No.2015R1A2A2A01007827
文摘AIM: To assess intravoxel incoherent motion diffusionweighted imaging(IVIM-DWI) for monitoring early efficacy of chemotherapy in a human gastric cancer mouse model.METHODS: IVIM-DWI was performed with 12 b-values(0-800 s/mm2) in 25 human gastric cancer-bearing nude mice at baseline(day 0), and then they were randomly divided into control and 1-, 3-, 5- and 7-d treatment groups(n = 5 per group). The control group underwent longitudinal MRI scans at days 1, 3, 5 and 7, and the treatment groups underwent subsequent MRI scans after a specified 5-fluorouracil/calciumfolinate treatment. Together with tumor volumes(TV), the apparent diffusion coefficient(ADC) and IVIM parameters [true water molecular diffusion coefficient(D), perfusion fraction(f) and pseudo-related diffusion coefficient(D*)] were measured. The differences in those parameters from baseline to each measurement(ΔTV%, ΔADC%, ΔD%, Δf% and ΔD*%) were calculated. After image acquisition, tumor necrosis, microvessel density(MVD) and cellular apoptosis were evaluated by hematoxylin-eosin(HE), CD31 and terminal-deoxynucleotidyl transferase mediated nick end labeling(TUNEL) staining respectively, to confirm the imaging findings. Mann-Whitney test and Spearman's correlation coefficient analysis were performed.RESULTS: The observed relative volume increase(ΔTV%) in the treatment group were significantly smaller than those in the control group at day 5(ΔTV_(treatment)% = 19.63% ± 3.01% and ΔTVcontrol% = 83.60% ± 14.87%, P = 0.008) and day 7(ΔTV_(treatment)% = 29.07% ± 10.01% and ΔTV_(control)% = 177.06% ± 63.00%, P = 0.008). The difference in ΔTV% between the treatment and the control groups was not significant at days 1 and 3 after a short duration of treatment. Increases in ADC in the treatment group(ΔADC%_(treatment), median, 30.10% ± 18.32%, 36.11% ± 21.82%, 45.22% ± 24.36%) were significantly higher compared with the control group(ΔADC%_(control), median, 4.98% ± 3.39%, 6.26% ± 3.08%, 9.24% ± 6.33%) at days 3, 5 and 7(P = 0.008, P = 0.016, P = 0.008, respectively). Increases in D in the treatment group(ΔD%_(treatment), median 17.12% ± 8.20%, 24.16% ± 16.87%, 38.54% ± 19.36%) were higher than those in the control group(ΔD%_(control), median-0.13% ± 4.23%, 5.89% ± 4.56%, 5.54% ± 4.44%) at days 1, 3, and 5(P = 0.032, P = 0.008, P = 0.016, respectively). Relative changes in f were significantly lower in the treatment group compared with the control group at days 1, 3, 5 and 7 follow-up(median,-34.13% ± 16.61% vs 1.68% ± 3.40%, P = 0.016;-50.64% ± 6.82% vs 3.01% ± 6.50%, P = 0.008;-49.93% ± 6.05% vs 0.97% ± 4.38%, P = 0.008, and-46.22% ± 7.75% vs 8.14% ± 6.75%, P = 0.008, respectively). D* in the treatment group decreased significantly compared to those in the control group at all time points(median,-32.10% ± 12.22% vs 1.85% ± 5.54%, P = 0.008;-44.14% ± 14.83% vs 2.29% ± 10.38%, P = 0.008;-59.06% ± 19.10% vs 3.86% ± 5.10%, P = 0.008 and-47.20% ± 20.48% vs 7.13% ± 9.88%, P = 0.016, respectively). Furthermore, histopathologic findings showed positive correlations with ADC and D and tumor necrosis(r_s = 0.720, P < 0.001; r_s = 0.522, P = 0.007, respectively). The cellular apoptosis of the tumor also showed positive correlations with ADC and D(r_s = 0.626, P = 0.001; r_s = 0.542, P = 0.005, respectively). Perfusionrelated parameters(f and D*) were positively correlated to MVD(r_s = 0.618, P = 0.001; r_s = 0.538, P = 0.006, respectively), and negatively correlated to cellular apoptosis of the tumor(r_s =-0.550, P = 0.004; r_s =-0.692, P < 0.001, respectively).CONCLUSION: IVIM-DWI is potentially useful for predicting the early efficacy of chemotherapy in a human gastric cancer mouse model.