针对奇异值分解(Singular value decomposition, SVD)的频率分离问题,研究了SVD对单个频率的分离条件,发现SVD分离单个频率的效果取决于各频率的幅值差异。若不同频率的幅值很接近,则SVD就不能分离这些频率,由此提出一种频率添加SVD算...针对奇异值分解(Singular value decomposition, SVD)的频率分离问题,研究了SVD对单个频率的分离条件,发现SVD分离单个频率的效果取决于各频率的幅值差异。若不同频率的幅值很接近,则SVD就不能分离这些频率,由此提出一种频率添加SVD算法。为了提取原信号中的特征频率,先对原信号添加该频率的理想正弦信号,使原信号中该频率成分和其他频率的幅值产生差异,从而实现对该频率成分的提取,从理论上证明此算法的可行性。仿真信号处理实例表明,即使对于频率值非常接近的两个频率,频率添加SVD算法亦可将它们准确分离,分离结果波形误差小,克服了原来SVD频率分离算法的缺陷。将此算法应用某转子系统的振动特征提取,准确地提取到振动的高阶倍频,发现高阶倍频振幅的周期性波动特征,并分析这种振幅周期性波动的原因。展开更多
Facial features under variant-expressions and partial occlusions could have degrading effect on overall face recognition performance. As a solution, we suggest that the contribution of these features on final classifi...Facial features under variant-expressions and partial occlusions could have degrading effect on overall face recognition performance. As a solution, we suggest that the contribution of these features on final classification should be determined. In order to represent facial features＇ contribution according to their variations, we propose a feature selection process that describes facial features as local independent component analysis （ICA） features. These local features are acquired using locally lateral subspace （LLS） strategy. Then, through linear discriminant analysis （LDA） we investigate the intraclass and interclass representation of each local ICA feature and express each feature＇s contribution via a weighting process. Using these weights, we define the contribution of each feature at local classifier level. In order to recognize faces under single sample constraint, we implement LLS strategy on locally linear embedding （LLE） along with the proposed feature selection. Additionally, we highlight the efficiency of the implementation of LLS strategy. The overall accuracy achieved by our approach on datasets with different facial expressions and partial occlusions such as AR, JAFFE, FERET and CK% is 90.70%. We present together in this paper survey results on face recognition performance and physiological feature selection performed by human subjects.展开更多
针对如何实现只有单训练样本情况下人脸认证,提出基于稀疏扩展字典学习的代价敏感单样本人脸认证方法。首先学习一种可将训练样本和一般训练集结合起来的投影方式来构造适合训练样本的稀疏扩展字典,而并非独立地利用一般训练集直接构造...针对如何实现只有单训练样本情况下人脸认证,提出基于稀疏扩展字典学习的代价敏感单样本人脸认证方法。首先学习一种可将训练样本和一般训练集结合起来的投影方式来构造适合训练样本的稀疏扩展字典,而并非独立地利用一般训练集直接构造扩展字典,从而更好地解决单训练样本不能涵盖测试条件变化的问题;其次通过稀疏表示分类得到与测试样本最相似的训练样本,然后对测试样本和该训练样本分别提取HOG特征,根据距离准则计算相似度判断是否在阈值范围内;最终实现在光照、表情变化情况下的单训练样本人脸鲁棒认证。该方法分别在AR、CMU-PIE和Extended Yale B 3个公共人脸数据库上进行实验,均取得较满意的结果,验证了该方法的可行性和有效性。展开更多
脉冲星搜寻是射电天文领域的重要研究方向。随着大型射电望远镜的不断建设和发展,数据量呈指数增长,如何及时从快速获取的海量数据中准确识别脉冲星信号成为当前面临的巨大挑战。以低频射电阵列(Low Frequency Array,LOFAR)联合阵列巡...脉冲星搜寻是射电天文领域的重要研究方向。随着大型射电望远镜的不断建设和发展,数据量呈指数增长,如何及时从快速获取的海量数据中准确识别脉冲星信号成为当前面临的巨大挑战。以低频射电阵列(Low Frequency Array,LOFAR)联合阵列巡天项目的观测数据为例,设计了针对单脉冲信号识别的10个特征变量,进一步研究了XGBoost结合包裹式特征选择法在单脉冲信号识别中的应用,并对比分析了GBDT(Gradient Boosting Decision Tree)、AdaBoost、随机森林和BP(Back Propagation)神经网络等模型对单脉冲信号识别的效果。实验结果表明,XGBoost结合包裹式特征选择法在单脉冲信号识别方面更具综合优势,误分类率最低,分类结果的精确率、召回率与F1分数最高,平均高出其他模型1到2个百分点。从特征选择上来说,有9个特征被选为最优特征。本研究设计的特征变量和识别方法可为我国开展以500 m口径球面射电望远镜(Five-hundred-meter Aperture Spherical radio Telescope,FAST)探测信号为主的脉冲星搜寻提供方法和技术参考。展开更多
In this study,we devised a computational framework called Supervised Feature Learning and Scoring(SuperFeat)which enables the training of a machine learning model and evaluates the canonical cellular statuses/features...In this study,we devised a computational framework called Supervised Feature Learning and Scoring(SuperFeat)which enables the training of a machine learning model and evaluates the canonical cellular statuses/features in pathological tissues that underlie the progression of disease.This framework also enables the identification of potential drugs that target the presumed detrimental cellular features.This framework was constructed on the basis of an artificial neural network with the gene expression profiles serving as input nodes.The training data comprised single-cell RNA sequencing datasets that encompassed the specific cell lineage during the developmental progression of cell features.A few models of the canonical cancer-involved cellular statuses/features were tested by such framework.Finally,we illustrated the drug repurposing pipeline,utilizing the training parameters derived from the adverse cellular statuses/features,which yielded successful validation results both in vitro and in vivo.SuperFeat is accessible at https://github.com/weilin-genomics/rSuperFeat.展开更多
文摘针对奇异值分解(Singular value decomposition, SVD)的频率分离问题,研究了SVD对单个频率的分离条件,发现SVD分离单个频率的效果取决于各频率的幅值差异。若不同频率的幅值很接近,则SVD就不能分离这些频率,由此提出一种频率添加SVD算法。为了提取原信号中的特征频率,先对原信号添加该频率的理想正弦信号,使原信号中该频率成分和其他频率的幅值产生差异,从而实现对该频率成分的提取,从理论上证明此算法的可行性。仿真信号处理实例表明,即使对于频率值非常接近的两个频率,频率添加SVD算法亦可将它们准确分离,分离结果波形误差小,克服了原来SVD频率分离算法的缺陷。将此算法应用某转子系统的振动特征提取,准确地提取到振动的高阶倍频,发现高阶倍频振幅的周期性波动特征,并分析这种振幅周期性波动的原因。
基金supported by Ministry of Higher Education MalaysiaUniversiti Teknologi MARA,Malaysia
文摘Facial features under variant-expressions and partial occlusions could have degrading effect on overall face recognition performance. As a solution, we suggest that the contribution of these features on final classification should be determined. In order to represent facial features＇ contribution according to their variations, we propose a feature selection process that describes facial features as local independent component analysis （ICA） features. These local features are acquired using locally lateral subspace （LLS） strategy. Then, through linear discriminant analysis （LDA） we investigate the intraclass and interclass representation of each local ICA feature and express each feature＇s contribution via a weighting process. Using these weights, we define the contribution of each feature at local classifier level. In order to recognize faces under single sample constraint, we implement LLS strategy on locally linear embedding （LLE） along with the proposed feature selection. Additionally, we highlight the efficiency of the implementation of LLS strategy. The overall accuracy achieved by our approach on datasets with different facial expressions and partial occlusions such as AR, JAFFE, FERET and CK% is 90.70%. We present together in this paper survey results on face recognition performance and physiological feature selection performed by human subjects.
文摘针对如何实现只有单训练样本情况下人脸认证,提出基于稀疏扩展字典学习的代价敏感单样本人脸认证方法。首先学习一种可将训练样本和一般训练集结合起来的投影方式来构造适合训练样本的稀疏扩展字典,而并非独立地利用一般训练集直接构造扩展字典,从而更好地解决单训练样本不能涵盖测试条件变化的问题;其次通过稀疏表示分类得到与测试样本最相似的训练样本,然后对测试样本和该训练样本分别提取HOG特征,根据距离准则计算相似度判断是否在阈值范围内;最终实现在光照、表情变化情况下的单训练样本人脸鲁棒认证。该方法分别在AR、CMU-PIE和Extended Yale B 3个公共人脸数据库上进行实验,均取得较满意的结果,验证了该方法的可行性和有效性。
基金supported by grants from the Shanghai Jiao Tong University,the Renji Hospital Start-up funding for New PI,the Natural Science Foundation of Shanghai Science and Technology Innovation Action Plan(Grant No.21ZR1441500)the Young Talent of Hunan(Grant No.2020RC3066)+1 种基金the Hunan Natural Science Fund for Excellent Young Scholars(Grant No.2021JJ20003)the China Postdoctoral Science Foundation(Grant No.2021T140197).
文摘In this study,we devised a computational framework called Supervised Feature Learning and Scoring(SuperFeat)which enables the training of a machine learning model and evaluates the canonical cellular statuses/features in pathological tissues that underlie the progression of disease.This framework also enables the identification of potential drugs that target the presumed detrimental cellular features.This framework was constructed on the basis of an artificial neural network with the gene expression profiles serving as input nodes.The training data comprised single-cell RNA sequencing datasets that encompassed the specific cell lineage during the developmental progression of cell features.A few models of the canonical cancer-involved cellular statuses/features were tested by such framework.Finally,we illustrated the drug repurposing pipeline,utilizing the training parameters derived from the adverse cellular statuses/features,which yielded successful validation results both in vitro and in vivo.SuperFeat is accessible at https://github.com/weilin-genomics/rSuperFeat.
