Some questions regarding the analysis of classifiers and classifier constructions are raised in this paper.The classifier,as a mere adjunct adjoining to the head,cannot serve as the head of the noun phrase containing ...Some questions regarding the analysis of classifiers and classifier constructions are raised in this paper.The classifier,as a mere adjunct adjoining to the head,cannot serve as the head of the noun phrase containing it,and as a result,it cannot project as ClP or nP.Under this approach,the DP analysis and the classifier construction theory are further refined.The constituents which precede and follow the classifier are analyzed in terms of their syntactic functions,semantic relations,linear features,feature assignment and syntactic occurrence in order to represent the classifier construction with the X-bar phrase structure theory appropriately and correctly and present a universal approach to classifier constructions in various languages.展开更多
A machine learning(ML)-based random forest(RF)classification model algorithm was employed to investigate the main factors affecting the formation of the core-shell structure of BaTiO_(3)-based ceramics and their inter...A machine learning(ML)-based random forest(RF)classification model algorithm was employed to investigate the main factors affecting the formation of the core-shell structure of BaTiO_(3)-based ceramics and their interpretability was analyzed by using Shapley additive explanations(SHAP).An F1-score changed from 0.8795 to 0.9310,accuracy from 0.8450 to 0.9070,precision from 0.8714 to 0.9000,recall from 0.8929 to 0.9643,and ROC/AUC value of 0.97±0.03 was achieved by the RF classification with the optimal set of features containing only 5 features,demonstrating the high accuracy of our model and its high robustness.During the interpretability analysis of the model,it was found that the electronegativity,melting point,and sintering temperature of the dopant contribute highly to the formation of the core-shell structure,and based on these characteristics,specific ranges were delineated and twelve elements were finally obtained that met all the requirements,namely Si,Sc,Mn,Fe,Co,Ni,Pd,Er,Tm,Lu,Pa,and Cm.In the process of exploring the structure of the core-shell,the doping elements can be effectively localized to be selected by choosing the range of features.展开更多
Design constraints verification is the most computationally expensive task in evolutionary structural optimization due to a large number of structural analyses thatmust be conducted.Building a surrogatemodel to approx...Design constraints verification is the most computationally expensive task in evolutionary structural optimization due to a large number of structural analyses thatmust be conducted.Building a surrogatemodel to approximate the behavior of structures instead of the exact structural analyses is a possible solution to tackle this problem.However,most existing surrogate models have been designed based on regression techniques.This paper proposes a novel method,called CaDE,which adopts a machine learning classification technique for enhancing the performance of the Differential Evolution(DE)optimization.The proposed method is separated into two stages.During the first optimization stage,the original DE is implemented as usual,but all individuals produced in this phase are stored as inputs of the training data.Based on design constraints verification,these individuals are labeled as“safe”or“unsafe”and their labels are saved as outputs of the training data.When collecting enough data,an AdaBoost model is trained to evaluate the safety state of structures.This model is then used in the second stage to preliminarily assess new individuals,and unpromising ones are rejected without checking design constraints.This method reduces unnecessary structural analyses,thereby shortens the optimization process.Five benchmark truss sizing optimization problems are solved using the proposed method to demonstrate its effectiveness.The obtained results show that the CaDE finds good optimal designs with less structural analyses in comparison with the original DE and four other DE variants.The reduction rate of five examples ranges from 18 to over 50%.Moreover,the proposed method is applied to a real-size transmission tower design problem to exhibit its applicability in practice.展开更多
Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类...Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类簇中心和局部密度可选出数据空间结构表现较好的样本。DPC-TT算法采用密度峰值聚类算法获取训练数据的类簇中心和样本的局部密度,对类簇中心的截断距离范围内的样本认定为空间结构表现较好,标记为核心数据,使用核心数据更新分类器,可降低迭代过程中的训练噪声,进而提高分类器的性能。实验结果表明:相比于标准Tritraining算法及其改进算法,DPC-TT算法具有更好的分类性能。展开更多
文摘Some questions regarding the analysis of classifiers and classifier constructions are raised in this paper.The classifier,as a mere adjunct adjoining to the head,cannot serve as the head of the noun phrase containing it,and as a result,it cannot project as ClP or nP.Under this approach,the DP analysis and the classifier construction theory are further refined.The constituents which precede and follow the classifier are analyzed in terms of their syntactic functions,semantic relations,linear features,feature assignment and syntactic occurrence in order to represent the classifier construction with the X-bar phrase structure theory appropriately and correctly and present a universal approach to classifier constructions in various languages.
基金Funded by the National Key Research and Development Program of China(No.2023YFB3812200)。
文摘A machine learning(ML)-based random forest(RF)classification model algorithm was employed to investigate the main factors affecting the formation of the core-shell structure of BaTiO_(3)-based ceramics and their interpretability was analyzed by using Shapley additive explanations(SHAP).An F1-score changed from 0.8795 to 0.9310,accuracy from 0.8450 to 0.9070,precision from 0.8714 to 0.9000,recall from 0.8929 to 0.9643,and ROC/AUC value of 0.97±0.03 was achieved by the RF classification with the optimal set of features containing only 5 features,demonstrating the high accuracy of our model and its high robustness.During the interpretability analysis of the model,it was found that the electronegativity,melting point,and sintering temperature of the dopant contribute highly to the formation of the core-shell structure,and based on these characteristics,specific ranges were delineated and twelve elements were finally obtained that met all the requirements,namely Si,Sc,Mn,Fe,Co,Ni,Pd,Er,Tm,Lu,Pa,and Cm.In the process of exploring the structure of the core-shell,the doping elements can be effectively localized to be selected by choosing the range of features.
基金funded by Hanoi University of Civil Engineering(HUCE)in Project Code 35-2021/KHXD-TD.
文摘Design constraints verification is the most computationally expensive task in evolutionary structural optimization due to a large number of structural analyses thatmust be conducted.Building a surrogatemodel to approximate the behavior of structures instead of the exact structural analyses is a possible solution to tackle this problem.However,most existing surrogate models have been designed based on regression techniques.This paper proposes a novel method,called CaDE,which adopts a machine learning classification technique for enhancing the performance of the Differential Evolution(DE)optimization.The proposed method is separated into two stages.During the first optimization stage,the original DE is implemented as usual,but all individuals produced in this phase are stored as inputs of the training data.Based on design constraints verification,these individuals are labeled as“safe”or“unsafe”and their labels are saved as outputs of the training data.When collecting enough data,an AdaBoost model is trained to evaluate the safety state of structures.This model is then used in the second stage to preliminarily assess new individuals,and unpromising ones are rejected without checking design constraints.This method reduces unnecessary structural analyses,thereby shortens the optimization process.Five benchmark truss sizing optimization problems are solved using the proposed method to demonstrate its effectiveness.The obtained results show that the CaDE finds good optimal designs with less structural analyses in comparison with the original DE and four other DE variants.The reduction rate of five examples ranges from 18 to over 50%.Moreover,the proposed method is applied to a real-size transmission tower design problem to exhibit its applicability in practice.
文摘Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类簇中心和局部密度可选出数据空间结构表现较好的样本。DPC-TT算法采用密度峰值聚类算法获取训练数据的类簇中心和样本的局部密度,对类簇中心的截断距离范围内的样本认定为空间结构表现较好,标记为核心数据,使用核心数据更新分类器,可降低迭代过程中的训练噪声,进而提高分类器的性能。实验结果表明:相比于标准Tritraining算法及其改进算法,DPC-TT算法具有更好的分类性能。
