检索结果-南通市图书馆

Learning-based counterfactual explanations for recommendation

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第8期67卷 115-129页

作者： jingxuan WEN Huafeng LIU liping jing Jian YU School of Computer and Information Technology Beijing Jiaotong University Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University

Counterfactual explanations provide explanations by exploring the changes in effect caused by changes in cause. They have attracted significant attention in recommender system research to explore the impact of changes in certain properties on the recommendation mechanism. Among several counterfactual recommendation methods, item-based counterfactual explanation methods have attracted considerable attention because of their flexibility. The core idea of item-based counterfactual explanation methods is to find a minimal subset of interacted items(i.e., short length) such that the recommended item would topple out of the top-K recommendation list once these items have been removed from user interactions(i.e., good quality). Usually, explanations are generated by ranking the precomputed importance of items, which fails to characterize the true importance of interacted items due to separation from the explanation generation. Additionally, the final explanations are generated according to a certain search strategy given the precomputed importance. This indicates that the quality and length of counterfactual explanations are deterministic;therefore, they cannot be balanced once the search strategy is fixed. To overcome these obstacles, this study proposes learning-based counterfactual explanations for recommendation(LCER) to provide counterfactual explanations based on personalized recommendations by jointly modeling the factual and counterfactual preference. To achieve consistency between the computation of importance and generation of counterfactual explanations, the proposed LCER endows an optimizable importance for each interacted item, which is supervised by the goal of counterfactual explanations to guarantee its credibility. Because of the model's flexibility, the trade-off between quality and length can be customized by setting different proportions. The experimental results on four real-world datasets demonstrate the effectiveness of the proposed LCER over several

关键词： recommender system explainable recommendation item-based explanation counterfactual inference counterfactual explanation

同方期刊数据库

Characteristics and factors that influenced damage to dams in the M_s 8.0 Wenchuan earthquake

在线全文

学校读者我要写书评

暂无评论

Earthquake Engineering and Engineering Vibration 2011年第3期10卷 349-358页

作者： jing liping Liang Haian Li Yongqiang Liu Chunhui Institute of Engineering Mechanics China Earthquake Administration Harbin 150080 China

Based on raw data from dams damaged in the Wenchuan earthquake, including many that were severely damaged, characteristics and factors that influenced the damage are discussed in this paper. Findings from this study include： severely damaged dams were densely distributed along the seismologic fault; small dams, especially small earth-rock dams, had the most serious damage that was caused by a variety of factors; the most serious damage was caused by seismic waves; damage was aggregated by aftershocks; and the extent of the damage patterns increased with the seismic intensity. Damage patterns varied in different intensity zones and cracking was the most common type of damage. Most of the dams had a good base with relatively high bearing capacity, and the walls of the earth-rock dams were mostly of clay soil. This type of base and body material mitigated some of the damage to dams. Reservoir maintenance and other factors also have a significant impact on the seismic safety of the dam. Finally, some recommendations to reduce seismic damage to dams are proposed.

关键词： seismic damage dam characteristic of damage to dams influencing factors Wenchuan earthquake seismologic fault earth-rock dams

Important sampling based active learning for imbalance classification

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2020年第8期63卷 196-209页

作者： Xinyue WANG Bo LIU Siyu CAO liping jing Jian YU School of Computer and Information Technology Beijing Jiaotong University Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University College of Information Science and Technology Hebei Agricultural University

Imbalance in data distribution hinders the learning performance of classifiers. To solve this problem, a popular type of methods is based on sampling(including oversampling for minority class and undersampling for majority class) so that the imbalanced data becomes relatively balanced data. However,they usually focus on one sampling technique, oversampling or undersampling. Such strategy makes the existing methods suffer from the large imbalance ratio(the majority instances size over the minority instances size). In this paper, an active learning framework is proposed to deal with imbalanced data by alternative performing important sampling(ALIS), which consists of selecting important majority-class instances and generating informative minority-class instances. In ALIS, two important sampling strategies affect each other so that the selected majority-class instances provide much clearer information in the next oversampling process, meanwhile the generated minority-class instances provide much more sufficient information for the next undersampling procedure. Extensive experiments have been conducted on real world datasets with a large range of imbalance ratio to verify ALIS. The experimental results demonstrate the superiority of ALIS in terms of several well-known evaluation metrics by comparing with the state-of-the-art methods.

关键词： imbalance classification important sampling active learning oversampling undersampling

Leveraging implicit social structures for recommendation via a Bayesian generative model

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2022年第4期65卷 269-271页

作者： Huafeng LIU jingxuan WEN liping jing Jian YU School of Computer and Information Technology Beijing Jiaotong University Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University

Dear editor,Social recommendation leverages the social relations between users and their past behaviors to model user preferences. While social recommendation is a core component in recommendation systems, discovering... 详细信息

关键词： Recommendation system Collaborative filtering Social recommendation Social networks Generative model

同方期刊数据库

土层地震反应计算选用模型与输入界面影响的研究

在线全文

学校读者我要写书评

暂无评论

地震科学进展 2025年第1期55卷 12-20页

作者：朱淑凯景立平齐文浩中国地震局工程力学研究所黑龙江哈尔滨150080

本文选用H-D模型和Davidenkov模型对土层动力试验本构参数进行拟合,讨论了本构模型对土层地震反应计算的影响;选取Ⅱ、Ⅲ、Ⅳ类场地各一个土层剖面,选定500 m/s剪切波速的界面作为地震动输入界面,下伏土层考虑为与输入土层性质一致,向下... 详细信息

本文选用H-D模型和Davidenkov模型对土层动力试验本构参数进行拟合,讨论了本构模型对土层地震反应计算的影响;选取Ⅱ、Ⅲ、Ⅳ类场地各一个土层剖面,选定500 m/s剪切波速的界面作为地震动输入界面,下伏土层考虑为与输入土层性质一致,向下取5~30 m的6个输入界面,选用Imperial Valley波、Northridge波以及Kobe作为输入地震动,用一维等效线性化方法进行计算,分析了输入界面选取对土层地震反应计算的影响。通过对结果的分析得出:①Davidenkov模型对于软土拟合优度优于H-D模型,但两者拟合得到的土层地震动反应区别较小,仅在0.40 g幅值输入情况下PGA表现出一定差别。②地表峰值加速度在下伏土层一定厚度内随着深度增加而增大;地表加速度反应谱下伏土层一定厚度内随着深度增加在短周期减小,在长周期和波峰增大。③土层地震反应分析在选定500 m/s剪切波速为钻孔终层和输入界面时,宜考虑5~20 m厚的与终层性质相同的下伏土层作为输入界面。

关键词：输入界面土的动力本构关系地表峰值加速度地表加速度反应谱

Siamese transformer with hierarchical concept embedding for fine-grained image recognition

同方期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2023年第3期66卷 188-203页

作者： Yilin LYU liping jing Jiaqi WANG Mingzhe GUO Xinyue WANG Jian YU School of Computer and Information Technology Beijing Jiaotong University Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University Alibaba Group

Distinguishing the subtle differences among fine-grained images from subordinate concepts of a concept hierarchy is a challenging task. In this paper, we propose a Siamese transformer with hierarchical concept embedding(STr HCE), which contains two transformer subnetworks sharing all configurations,and each subnetwork is equipped with the hierarchical semantic information at different concept levels for fine-grained image embeddings. In particular, one subnetwork is for coarse-scale patches to learn the discriminative regions with the aid of the innate multi-head self-attention mechanism of the transformer. The other subnetwork is for finer-scale patches, which are adaptively sampled from the discriminative regions, to capture subtle yet discriminative visual cues and eliminate redundant information. STr HCE connects the two subnetworks through a score margin adjustor to enforce the most discriminative regions generating more confident predictions. Extensive experiments conducted on four commonly-used benchmark datasets, including CUB-200-2011, FGVC-Aircraft, Stanford Dogs, and NABirds, empirically demonstrate the superiority of the proposed STr HCE over state-of-the-art baselines.

关键词： fine-grained image recognition transformer hierarchical concept embedding adaptive sampling Siamese network

Understanding the defect chemistry of oxide nanoparticles for creating new functionalities:A critical review

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China Chemistry 2011年第6期54卷 876-886页

作者： LI GuangShe,LI liping & ZHENG jing State Key Laboratory of Structural Chemistry Fujian Institute of Research on the Structure of Matter,Chinese Academy of Sciences,Fuzhou 350002,China State Key Laboratory of Structural Chemistry Fujian Institute of Research on the Structure of Matter Chinese Academy of Sciences Fuzhou China

This work presents a critical review on the studies of defect chemistry of oxide nanoparticles for creating new functionalities pertinent to energy applications including dilute-magnetic semiconductors,giant-dielectrics,or white light *** is placed on the relationships between the internal structure and defective surfaces of oxide nanoparticles and their synergy in tailoring the materials *** review is arranged in a sequence:(1) structural fundamentals of bulk oxides,using TiO2 as a model simple oxide to highlight the importance of polymorphs in tuning the electronic structures;(2) structural features of simple oxide nanoparticles distinct from the bulk,which show that nanoparticles can be considered as a special solid under the compression as originated from the surface defect dipole-dipole interactions;and(3) new functions achieved through extending the defect chemistry concept to the assembled architectures or multi-component oxide nanoparticles,in which defect surfaces enable the localized electrons or intermediate levels to produce giant dielectric performance or tunable light *** is concluded that understandings of defect chemistry provide diverse possibilities to manipulate electrons in oxide nanoparticles for functionalities in energy-relevant applications.

关键词： nanoparticles defect chemistry surfaces functionalities

Promoter Hypomethylation and Increased Expression of Syncytin-1 in Non-small Cell Lung Cancer

在线全文

学校读者我要写书评

暂无评论

Promoter Hypomethylation and Increased Expression of Syncyti...

2021年中国肿瘤标志物学术大会暨第十五届肿瘤标志物青年科学家论坛

作者： liping jing Xuewei Zhuang Third Shandong Provincial Hospital

Purpose: In this study, fluctuation of syncytin-1 expression in non-small cell lung cancer(NSCLC) and relative para-cancerous tissues during its prognosis as a plausible pathological mechanism of NSCLC is discussed by statistical analysis of syncytin-1 gene methylation and expression ***: Syncytin-1 expressions in NSCLC and its relative para-cancerous tissues were verified by immunohistochemistry method. Differentiation of between 5-year survival and death group was analyzed. Survival ratio was calculated by Kaplan-Meier survival curve. Death risk assessment was executed by Cox risk regression model. 5-LTR methylation level of syncytin-1 promoter was detected by EpiTYPER ***: Syncytin-1 expression in NSCLC tissue is higher rather than para-cancerous tissues in significant differences with P<0.01 and in 5-year survival group is lower than death in significant difference with P<0.01 respectively. Based on the average survival period of higher syncytin-1 expression group is significantly lower than the lower one with P<0.01, clinical stage and syncytin-1 positive ratio were considered as top risk factors according to Cox ratio risk regression model analysis. The total methylation ratio of NSCLC tissues is lower than para-carcinoma tissues with no significant difference but the CpG-2 site methylation ratio of NSCLC tissue is lower than para-carcinoma tissues in significant difference with P<***: The low expression level of syncytin-1 is qualified to be an important prognosis *** of the syncytin-1 promoter region in non-small cell lung cancer promotes the expression of its gene.

关键词： Non-small cell lung cancer Syncytin-1 expression level prognosis methylation

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

基于神经网络探究标签依赖关系的多标签分类

计算机研究与发展 2018年第8期55卷 1751-1759页

作者：宋攀景丽萍交通数据分析与挖掘北京市重点实验室(北京交通大学) 北京100044

多标签学习广泛应用于文本分类、图像标注、视频语义注释、基因功能分析等问题.最近,多标签学习受到大量的关注,成为机器学习领域中的研究热点.然而,已有的算法并不能充分地探究标签之间的依赖关系和解决标签缺失问题,为此提出一种基于... 详细信息

多标签学习广泛应用于文本分类、图像标注、视频语义注释、基因功能分析等问题.最近,多标签学习受到大量的关注,成为机器学习领域中的研究热点.然而,已有的算法并不能充分地探究标签之间的依赖关系和解决标签缺失问题,为此提出一种基于神经网络探究标签依赖关系的算法NN_AD_Omega,它能够有效地处理这2个挑战.NN_AD_Omega算法在神经网络顶层加入Ω矩阵刻画标签之间的依赖关系,标签之间的依赖关系可通过充分挖掘数据内在特点得到.当实例部分标签缺失时,学到的标签之间依赖关系能够有效提高预测效果.为了高效地求解模型,采用最小批梯度下降方法(Mini-batch-GD),其中学习率的自适应计算采用AdaGrad技术.在4个标准多标签数据集上的实验结果表明,提出的算法能够探究标签之间的依赖关系和处理标签缺失问题,且其性能优于当前基于神经网络的多标签学习算法.

关键词：神经网络标签依赖关系多标签学习分类标签缺失