针对现有序列推荐模型忽略了不同用户的个性化行为,导致模型不能充分捕获用户动态偏好而产生的兴趣漂移等问题,提出了一种基于用户长短期偏好的序列推荐模型(Sequential Recommendation Model Based on User’s Long and Short Term Pre...针对现有序列推荐模型忽略了不同用户的个性化行为,导致模型不能充分捕获用户动态偏好而产生的兴趣漂移等问题,提出了一种基于用户长短期偏好的序列推荐模型(Sequential Recommendation Model Based on User’s Long and Short Term Preference,ULSP-SRM)。首先,根据用户的序列中交互物品的类别和时间信息生成用户的动态类别嵌入,进而有效建立物品之间的关联性,并且降低数据的稀疏性;其次,根据用户当前点击物品和最后一项点击的时间间隔信息生成个性化时序位置嵌入矩阵,模拟用户的个性化聚集现象,以更好地反映用户偏好的动态变化;然后,将融合了个性化时序位置嵌入矩阵的用户长期偏好序列以会话为单位输入门控循环单元中,生成用户的长期偏好表示,并通过注意力机制将用户长短期偏好进行融合,生成用户的最终偏好表示,从而达到充分捕获用户偏好的目的;最后,将用户最终偏好表示输入推荐预测层进行下一项推荐预测。在Amazon公开数据集的7个子集上进行实验,采用AUC(Area Under Curve)值、召回率和精确率指标进行综合评估,实验结果表明,所提模型的表现优于其他先进基准模型,有效地提升了推荐性能。展开更多
By analyzing the variability of global SST （sea surface temperature） anomalies, we propose a unified Nifio index using the surface thermal centroid anomaly of the region along the Pacific equator embraced by the 0.7...By analyzing the variability of global SST （sea surface temperature） anomalies, we propose a unified Nifio index using the surface thermal centroid anomaly of the region along the Pacific equator embraced by the 0.7~C contour line of the standard deviation of the SST anomalies and try to unify the traditional Nifio regions into a single entity. The unified Nifio region covers almost all of the traditional Nifio regions. The anomaly time series of the averaged SST over this region are closely correlated to historical Nifio indices. The anomaly time series of the zonal and meridional thermal centroid have close correlation with historical TNI （Trans-Nifio index） indices, showing differences among E1 Nifio （La Nifia） events. The meridional centroid anomaly suggests that areas of maximum temperature anomaly are moving meridionally （although slightly） with synchronous zonal movement. The zonal centroid anomalies of the unified Nifio region are found helpful in the classification of the Eastern Pacific （EP）/Central Pacific （CP） types of E1 Nifio events. More importantly, the zonal centroid anomaly shows that warm areas might move during a single warming/cooling phase. All the current Nifio indices can be well represented by a simple linear combination of unified Nifio indices, which suggests that the thermal anomaly （SSTA） and thermal centroid location anomaly of the unified Nifio region would yield a more complete image of each E1 Nifio/ La Nina event.展开更多
The threshold GARCH(TGARCH)models have been very useful for analyzing asymmetric volatilities arising from financial time series.Most research on TGARCH has been directed to the stationary case.This paper studies the ...The threshold GARCH(TGARCH)models have been very useful for analyzing asymmetric volatilities arising from financial time series.Most research on TGARCH has been directed to the stationary case.This paper studies the estimation of non-stationary first order TGARCH models.Restricted normal mixture quasi-maximum likelihood estimation(NM-QMLE)for non-stationary TGARCH models is proposed in the sense that we estimate the other parameters with any fixed location parameter.We show that the proposed estimators(except location parameter)are consistent and asymptotically normal under mild regular conditions.The impact of relative leptokursis and skewness of the innovations’distribution and quasi-likelihood distributions on the asymptotic efficiency has been discussed.Numerical results lend further support to our theoretical results.Finally,an illustrated real example is presented.展开更多
The maximum frequency of gravitational waves(GWs) detectable with traditional pulsar timing methods is set by the Nyquist frequency( fNy) of the observation. Beyond this frequency, GWs leave no temporal-correlated sig...The maximum frequency of gravitational waves(GWs) detectable with traditional pulsar timing methods is set by the Nyquist frequency( fNy) of the observation. Beyond this frequency, GWs leave no temporal-correlated signals; instead, they appear as white noise in the timing residuals. The variance of the GW-induced white noise is a function of the position of the pulsars relative to the GW source. By observing this unique functional form in the timing data, we propose that we can detect GWs of frequency >f_(Ny)(super-Nyquist frequency GWs; SNFGWs). We demonstrate the feasibility of the proposed method with simulated timing data.Using a selected dataset from the Parkes Pulsar Timing Array data release 1 and the North American Nanohertz Observatory for Gravitational Waves publicly available datasets, we try to detect the signals from single SNFGW sources. The result is consistent with no GW detection with 65.5% probability. An all-sky map of the sensitivity of the selected pulsar timing array to single SNFGW sources is generated, and the position of the GW source where the selected pulsar timing array is most sensitive to is λ_s =.0.82,β_s =-1.03(rad); the corresponding minimum GW strain is h = 6.31 × 10^(-11) at f = 1 × 10^(-5) Hz.展开更多
We propose that locations of genes on chromosomes can contribute to the prediction of gene regulatory relationships. We constructed a time-based gene regulatory network of zebrafish cardiogenesis on the basis of a spa...We propose that locations of genes on chromosomes can contribute to the prediction of gene regulatory relationships. We constructed a time-based gene regulatory network of zebrafish cardiogenesis on the basis of a spatio-temporal neighborhood method. Through the network, specific regulatory pathways and order of gene expression during zebrafish cardiogenesis were obtained. By comparing the order with locations of these genes on chromosomes, we discovered that there exists a reversal phenomenon between the order and order of gene locations. The discovery provides an inherent rule to instruct exploration of gene regulatory relationships. Specifically, the discovery can help to predict if regulatory relationships between genes exist and contribute to evaluating the correctness of discovered gene regulatory relationships.展开更多
文摘针对现有序列推荐模型忽略了不同用户的个性化行为,导致模型不能充分捕获用户动态偏好而产生的兴趣漂移等问题,提出了一种基于用户长短期偏好的序列推荐模型(Sequential Recommendation Model Based on User’s Long and Short Term Preference,ULSP-SRM)。首先,根据用户的序列中交互物品的类别和时间信息生成用户的动态类别嵌入,进而有效建立物品之间的关联性,并且降低数据的稀疏性;其次,根据用户当前点击物品和最后一项点击的时间间隔信息生成个性化时序位置嵌入矩阵,模拟用户的个性化聚集现象,以更好地反映用户偏好的动态变化;然后,将融合了个性化时序位置嵌入矩阵的用户长期偏好序列以会话为单位输入门控循环单元中,生成用户的长期偏好表示,并通过注意力机制将用户长短期偏好进行融合,生成用户的最终偏好表示,从而达到充分捕获用户偏好的目的;最后,将用户最终偏好表示输入推荐预测层进行下一项推荐预测。在Amazon公开数据集的7个子集上进行实验,采用AUC(Area Under Curve)值、召回率和精确率指标进行综合评估,实验结果表明,所提模型的表现优于其他先进基准模型,有效地提升了推荐性能。
基金Supported by the National Basic Research Program of China(973 Program)(Nos.2012CB957704,2009CB723903)the National Natural Science Foundation of China(Nos.40506035,40876005)
文摘By analyzing the variability of global SST （sea surface temperature） anomalies, we propose a unified Nifio index using the surface thermal centroid anomaly of the region along the Pacific equator embraced by the 0.7~C contour line of the standard deviation of the SST anomalies and try to unify the traditional Nifio regions into a single entity. The unified Nifio region covers almost all of the traditional Nifio regions. The anomaly time series of the averaged SST over this region are closely correlated to historical Nifio indices. The anomaly time series of the zonal and meridional thermal centroid have close correlation with historical TNI （Trans-Nifio index） indices, showing differences among E1 Nifio （La Nifia） events. The meridional centroid anomaly suggests that areas of maximum temperature anomaly are moving meridionally （although slightly） with synchronous zonal movement. The zonal centroid anomalies of the unified Nifio region are found helpful in the classification of the Eastern Pacific （EP）/Central Pacific （CP） types of E1 Nifio events. More importantly, the zonal centroid anomaly shows that warm areas might move during a single warming/cooling phase. All the current Nifio indices can be well represented by a simple linear combination of unified Nifio indices, which suggests that the thermal anomaly （SSTA） and thermal centroid location anomaly of the unified Nifio region would yield a more complete image of each E1 Nifio/ La Nina event.
基金supported by National Natural Science Foundation of China (Grant No.11101448)the Program for New Century Excellent Talents in University+3 种基金the Program for Young Talents of Beijing (Grant No.YETP0955)the Program for National Statistics Science Research Plan (Grant No.2013LY015)the "Project 211" of the Central University of Finance and Economicsthe Central University of Finance Young Scholar Innovation Fund
文摘The threshold GARCH(TGARCH)models have been very useful for analyzing asymmetric volatilities arising from financial time series.Most research on TGARCH has been directed to the stationary case.This paper studies the estimation of non-stationary first order TGARCH models.Restricted normal mixture quasi-maximum likelihood estimation(NM-QMLE)for non-stationary TGARCH models is proposed in the sense that we estimate the other parameters with any fixed location parameter.We show that the proposed estimators(except location parameter)are consistent and asymptotically normal under mild regular conditions.The impact of relative leptokursis and skewness of the innovations’distribution and quasi-likelihood distributions on the asymptotic efficiency has been discussed.Numerical results lend further support to our theoretical results.Finally,an illustrated real example is presented.
基金supported by the National Basic Research Program of China(Grant Nos.2014CB845802 and 2012CB821801)the National Natural Science Foundation of China(Grant Nos.11103019,11133002,11103022 and11373036)+1 种基金the Qianren Start-up Grant(Grant No.292012312D1117210)the Strategic Priority Research Program “The Emergence of Cosmological Structures”(Grant No.XDB09000000) of the Chinese Academy of Sciences
文摘The maximum frequency of gravitational waves(GWs) detectable with traditional pulsar timing methods is set by the Nyquist frequency( fNy) of the observation. Beyond this frequency, GWs leave no temporal-correlated signals; instead, they appear as white noise in the timing residuals. The variance of the GW-induced white noise is a function of the position of the pulsars relative to the GW source. By observing this unique functional form in the timing data, we propose that we can detect GWs of frequency >f_(Ny)(super-Nyquist frequency GWs; SNFGWs). We demonstrate the feasibility of the proposed method with simulated timing data.Using a selected dataset from the Parkes Pulsar Timing Array data release 1 and the North American Nanohertz Observatory for Gravitational Waves publicly available datasets, we try to detect the signals from single SNFGW sources. The result is consistent with no GW detection with 65.5% probability. An all-sky map of the sensitivity of the selected pulsar timing array to single SNFGW sources is generated, and the position of the GW source where the selected pulsar timing array is most sensitive to is λ_s =.0.82,β_s =-1.03(rad); the corresponding minimum GW strain is h = 6.31 × 10^(-11) at f = 1 × 10^(-5) Hz.
基金Project supported by the National Natural Science Foundation of China(Nos.60574079,30872957,and 30572129)
文摘We propose that locations of genes on chromosomes can contribute to the prediction of gene regulatory relationships. We constructed a time-based gene regulatory network of zebrafish cardiogenesis on the basis of a spatio-temporal neighborhood method. Through the network, specific regulatory pathways and order of gene expression during zebrafish cardiogenesis were obtained. By comparing the order with locations of these genes on chromosomes, we discovered that there exists a reversal phenomenon between the order and order of gene locations. The discovery provides an inherent rule to instruct exploration of gene regulatory relationships. Specifically, the discovery can help to predict if regulatory relationships between genes exist and contribute to evaluating the correctness of discovered gene regulatory relationships.
