Study of physisorbed and chemisorbed carbon dioxide (CO<sub>2</sub>) species was carried out on the NaX zeolite modified by cationic exchanges with bivalent cations (Ca<sup>2+</sup> and Ba<s...Study of physisorbed and chemisorbed carbon dioxide (CO<sub>2</sub>) species was carried out on the NaX zeolite modified by cationic exchanges with bivalent cations (Ca<sup>2+</sup> and Ba<sup>2+</sup>) by temperature-programmed desorption of CO<sub>2</sub> (CO<sub>2</sub>-TPD). Others results were obtained by infrared to complete the study. The results of this research showed, in the physisorption region (213 - 473 K), that the cationic exchanges on NaX zeolite with bivalent cations increase slightly the interactions of CO<sub>2</sub> molecule with adsorbents and/or cationic site. Indeed, the desorption energies of physisorbed CO<sub>2</sub> obtained on the reference zeolite NaX (13.5 kJ·mol<sup>-1</sup>) are lower than that of exchanged zeolites E-CaX and E-BaX (15.77 and 15.17 kJ·mol<sup>-1</sup> respectively). In the chemisorbed CO<sub>2</sub> region (573 - 873 K), the desorption energies related to desorbed species (bidentate carbonates: CO<sub>3</sub>2-</sup>) on the exchanged zeolites E-CaX and E-BaX are about 81 kJ·mol<sup>-1</sup>, higher than the desorbed species (bicarbonates: HCO<sub>3</sub>2-</sup>) on the reference R-NaX (62 kJ·mol<sup>-1</sup>). In addition, the exchanged E-BaX zeolite develops the secondary adsorption sites corresponding to bicarbonates species with desorption energies of 35 kJ·mol<sup>-1</sup> lower to desorption energies of bicarbonates noted on the reference zeolite NaX.展开更多
考虑将特征选择集成到支持向量机分类器中,提出集成特征选择的最优化支持向量机分类器——FS-SDPSVM(Feature Selection in Semi-definite Program for Support Vector Machine)。该模型将每个特征分别在核空间中做特征映射,然后通过参...考虑将特征选择集成到支持向量机分类器中,提出集成特征选择的最优化支持向量机分类器——FS-SDPSVM(Feature Selection in Semi-definite Program for Support Vector Machine)。该模型将每个特征分别在核空间中做特征映射,然后通过参数组合构成新的核矩阵,将特征选择过程与机器分类过程统一在一个优化目标下,同时达到特征选择与分类最优。在特征筛选方面,根据模型参数提出用于特征筛选的特征支持度和特征贡献度,通过控制二者的上下限可以在最优分类和最少特征之间灵活取舍。实证中分别将最优分类(FS-SDP-SVM1)和最少特征(FS-SDPSVM2)两类集成化特征选择算法与Relief-F、SFS、SBS算法在UCI机器学习数据和人造数据中进行对比实验。结果表明,提出的FS-SDP-SVM算法在保持较好泛化能力的基础上,在多数实验数据集中实现了最大分类准确率或最少特征数量;在人工数据中,该方法可以准确地选出真正的特征,去除噪声特征。展开更多
文摘Study of physisorbed and chemisorbed carbon dioxide (CO<sub>2</sub>) species was carried out on the NaX zeolite modified by cationic exchanges with bivalent cations (Ca<sup>2+</sup> and Ba<sup>2+</sup>) by temperature-programmed desorption of CO<sub>2</sub> (CO<sub>2</sub>-TPD). Others results were obtained by infrared to complete the study. The results of this research showed, in the physisorption region (213 - 473 K), that the cationic exchanges on NaX zeolite with bivalent cations increase slightly the interactions of CO<sub>2</sub> molecule with adsorbents and/or cationic site. Indeed, the desorption energies of physisorbed CO<sub>2</sub> obtained on the reference zeolite NaX (13.5 kJ·mol<sup>-1</sup>) are lower than that of exchanged zeolites E-CaX and E-BaX (15.77 and 15.17 kJ·mol<sup>-1</sup> respectively). In the chemisorbed CO<sub>2</sub> region (573 - 873 K), the desorption energies related to desorbed species (bidentate carbonates: CO<sub>3</sub>2-</sup>) on the exchanged zeolites E-CaX and E-BaX are about 81 kJ·mol<sup>-1</sup>, higher than the desorbed species (bicarbonates: HCO<sub>3</sub>2-</sup>) on the reference R-NaX (62 kJ·mol<sup>-1</sup>). In addition, the exchanged E-BaX zeolite develops the secondary adsorption sites corresponding to bicarbonates species with desorption energies of 35 kJ·mol<sup>-1</sup> lower to desorption energies of bicarbonates noted on the reference zeolite NaX.
文摘考虑将特征选择集成到支持向量机分类器中,提出集成特征选择的最优化支持向量机分类器——FS-SDPSVM(Feature Selection in Semi-definite Program for Support Vector Machine)。该模型将每个特征分别在核空间中做特征映射,然后通过参数组合构成新的核矩阵,将特征选择过程与机器分类过程统一在一个优化目标下,同时达到特征选择与分类最优。在特征筛选方面,根据模型参数提出用于特征筛选的特征支持度和特征贡献度,通过控制二者的上下限可以在最优分类和最少特征之间灵活取舍。实证中分别将最优分类(FS-SDP-SVM1)和最少特征(FS-SDPSVM2)两类集成化特征选择算法与Relief-F、SFS、SBS算法在UCI机器学习数据和人造数据中进行对比实验。结果表明,提出的FS-SDP-SVM算法在保持较好泛化能力的基础上,在多数实验数据集中实现了最大分类准确率或最少特征数量;在人工数据中,该方法可以准确地选出真正的特征,去除噪声特征。
