检索结果-内蒙古大学图书馆

Class-Imbalanced Semi-Supervised Learning for Large-Scale Point Cloud Semantic Segmentation via Decoupling Optimization

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Li, Mengtian Lin, Shaohui Wang, Zihan Shen, Yunhang Zhang, Baochang Ma, Lizhuang Shanghai University China Shanghai Engineering Research Center of Motion Picture Special Effects China East China Normal University China Key Laboratory of Advanced Theory and Application in Statistics and Data Science Ministry of Education China Beihang University China Tencent Youtu Lab China

Semi-supervised learning (SSL), thanks to the significant reduction of data annotation costs, has been an active research topic for large-scale 3D scene understanding. However, the existing SSL-based methods suffer from severe training bias, mainly due to class imbalance and long-tail distributions of the point cloud data. As a result, they lead to a biased prediction for the tail class segmentation. In this paper, we introduce a new decoupling optimization framework, which disentangles feature representation learning and classifier in an alternative optimization manner to shift the bias decision boundary effectively. In particular, we first employ two-round pseudo-label generation to select unlabeled points across head-to-tail classes. We further introduce multi-class imbalanced focus loss to adaptively pay more attention to feature learning across head-to-tail classes. We fix the backbone parameters after feature learning and retrain the classifier using ground-truth points to update its parameters. Extensive experiments demonstrate the effectiveness of our method outperforming previous state-of-the-art methods on both indoor and outdoor 3D point cloud datasets (i.e., S3DIS, ScanNet-V2, Semantic3D, and SemanticKITTI) using 1% and 1pt evaluation. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

ENSEMBLE DOMAIN DECOMPOSITION ALGORITHM FOR THE FULLY-MIXED RANDOM STOKES-DARCY MODEL WITH THE BEAVERS-JOSEPH INTERFACE CONDITIONS

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Shi, Feng Sun, Yizhong Zheng, Haibiao College of Science Harbin Institute of Technology Shenzhen China School of Mathematical Sciences East China Normal University Shanghai China School of Mathematical Sciences East China Normal University Shanghai Key Laboratory of Pure Mathematics and Mathematical Practice Key Laboratory of Advanced Theory and Application in Statistics and Data Science East China Normal University Shanghai China

In this paper, an efficient ensemble domain decomposition algorithm is proposed for fast solving the fully-mixed random Stokes-Darcy model with the physically realistic Beavers-Joseph (BJ) interface conditions. We utilize the Monte Carlo method for the coupled model with random inputs to derive some deterministic Stokes-Darcy numerical models and use the idea of the ensemble to realize the fast computation of multiple problems. One remarkable feature of the algorithm is that multiple linear systems share a common coefficient matrix in each deterministic numerical model, which significantly reduces the computational cost and achieves comparable accuracy with the traditional methods. Moreover, by domain decomposition, we can decouple the Stokes-Darcy system into two smaller sub-physics problems naturally. Both mesh-dependent and mesh-independent convergence rates of the algorithm are rigorously derived by choosing suitable Robin parameters. Optimized Robin parameters are derived and analyzed to accelerate the convergence of the proposed algorithm. Especially, for small hydraulic conductivity in practice, the almost optimal geometric convergence can be obtained by finite element discretization. Finally, two groups of numerical experiments are conducted to validate the exclusive features of the proposed algorithm. © 2022, CC BY-NC-ND.

关键词： Numerical models

On the estimation of high-dimensional integrated covariance matrices based on high-frequency data with multiple transactions

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Wang, Moming Xia, Ningning Zhou, Yong School of Statistics and Management Shanghai University of Finance and Economics 777 Guo Ding Road Shanghai200433 China School of Statistics and Management Shanghai Key Laboratory of Financial Information Technology Shanghai University of Finance and Economics 777 Guo Ding Road Shanghai200433 China Faculty of Economics and Management East China Normal University Shanghai200241 China Academy of Mathematics and Systems Science Chinese Academy of Science Beijing100190 China Key Laboratory of Advanced Theory and Application in Statistics and Data Science MOE Academy of Statistics and Interdisciplinary Sciences School of Statistics East China Normal University Shanghai200062 China

High-frequency data in financial markets often include multiple transactions at each recording time due to the mechanism of recording. Using random matrix theory, this paper considers the estimation of integrated covariance (ICV) matrices of high-dimensional diffusion processes based on multiple high-frequency observations. We start by studying the estimator, the time-variation adjusted realized covariance (TVA) matrix proposed by Zheng and Li (2011), without microstructure noise. We show that in the high-dimensional case, for a class C of diffusion processes, the limiting spectral distribution (LSD) of the averaged TVA depends not only on that of the ICV but also on the number of multiple transactions at each recording time. However, in practice, observed prices are always contaminated by market microstructure noise. Thus, we also study the limiting behavior of pre-averaging averaged TVA matrices based on noisy multiple observations. We show that for processes in class C, the pre-averaging averaged TVA has two desirable properties: it eliminates the effects of microstructure noise and multiple transactions, and its LSD depends solely on that of the ICV matrix. Further, three types of nonlinear shrinkage estimators of the ICV matrix are proposed based on high-frequency noisy multiple observations. Simulation studies support our theoretical results and demonstrate the finite sample performance of the proposed estimators. Finally, high-frequency portfolio strategies are evaluated under these estimators in real data analysis. Copyright © 2019, The Authors. All rights reserved.

关键词： Covariance matrix

On stochastic control problems with higher-order moments

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Yike Liu, Jingzhen Bensoussan, Alain Yiu, Ka-Fai Cedric Wei, Jiaqin School of Finance Chongqing Technology and Business University Chongqing400067 China China Institute for Actuarial Science Central University of Finance and Economics Beijing100081 China Naveen Jindal School of Management University of Texas at Dallas DallasTX75083 United States Department of Applied Mathematics The Hong Kong Polytechnic University Kowloon Hunghom Hong Kong Key Laboratory of Advanced Theory and Application in Statistics and Data Science MOE School of Statistics East China Normal University Shanghai200241 China

In this paper, we focus on a class of time-inconsistent stochastic control problems, where the objective function includes the mean and several higher-order central moments of the terminal value of state. To tackle the time-inconsistency, we seek both the closed-loop and the open-loop Nash equilibrium controls as time-consistent solutions. We establish a partial differential equation (PDE) system for deriving a closed-loop Nash equilibrium control, which does not include the equilibrium value function and is different from the extended Hamilton-Jacobi-Bellman (HJB) equations as in Björk et al. (Finance Stoch. 21: 331-360, 2017). We show that our PDE system is equivalent to the extended HJB equations that seems difficult to be solved for our higher-order moment problems. In deriving an open-loop Nash equilibrium control, due to the non-separable higher-order moments in the objective function, we make some moment estimates in addition to the standard perturbation argument for developing a maximum principle. Then, the problem is reduced to solving a flow of forward-backward stochastic differential equations. In particular, we investigate linear controlled dynamics and some objective functions affine in the mean. The closed-loop and the open-loop Nash equilibrium controls are identical, which are independent of the state value, random path and the preference on the odd-order central moments. By sending the highest order of central moments to infinity, we obtain the time-consistent solutions to some control problems whose objective functions include some penalty functions for *** Codes Primary: 93E20, 91G80, Secondary: 91B08, 49N90 Copyright © 2024, The Authors. All rights reserved.

关键词： Nash equilibrium

Measurement of solar pp neutrino flux using electron recoil data from PandaX-4T commissioning run

学校读者我要写书评

暂无评论

Chinese Physics C 2024年第9期48卷 1-7,300页

作者： Xiaoying Lu Abdusalam Abdukerim Zihao Bo Wei Chen Xun Chen Yunhua Chen Chen Cheng Zhaokan Cheng Xiangyi Cui Yingjie Fan Deqing Fang Lisheng Geng Karl Giboni Xuyuan Guo Chencheng Han Ke Han Changda He Jinrong He Di Huang Junting Huang Zhou Huang Ruquan Hou Yu Hou Xiangdong Ji Yonglin Ju Chenxiang Li Jiafu Li Mingchuan Li Shuaijie Li Tao Li Qing Lin Jianglai Liu Congcong Lu Lingyin Luo Yunyang Luo Wenbo Ma Yugang Ma Yajun Mao Yue Meng Xuyang Ning Binyu Pang Ningchun Qi Zhicheng Qian Xiangxiang Ren Nasir Shaheed Xiaofeng Shang Xiyuan Shao Guofang Shen Manbin Shen Lin Si Wenliang Sun Yi Tao Anqing Wang Meng Wang Qiuhong Wang Shaobo Wang Siguang Wang Wei Wang Xiuli Wang Xu Wang Zhou Wang Yuehuan Wei Mengmeng Wu Weihao Wu Yuan Wu Mengjiao Xiao Xiang Xiao Kaizhi Xiong Binbin Yan Xiyu Yan Yong Yang Chunxu Yu Ying Yuan Zhe Yuan Youhui Yun Xinning Zeng Minzhen Zhang Peng Zhang Shibo Zhang Shu Zhang Tao Zhang Wei Zhang Yang Zhang Yingxin Zhang Yuanyuan Zhang Li Zhao Jifang Zhou Ning Zhou Xiaopeng Zhou Yubo Zhou Zhizhen Zhou Research Center for Particle Science and Technology Institute of Frontier and Interdisciplinary ScienceShandong UniversityQingdao 266237China Key Laboratory of Particle Physics and Particle Irradiation of Ministry of Education Shandong UniversityQingdao 266237China School of Physics and Astronomy Shanghai Jiao Tong UniversityKey Laboratory for Particle Astrophysics and Cosmology(MoE)Shanghai Key Laboratory for Particle Physics and CosmologyShanghai 200240China New Cornerstone Science Laboratory Tsung-Dao Lee InstituteShanghai Jiao Tong UniversityShanghai 200240China Shanghai Jiao Tong University Sichuan Research Institute Chengdu 610213China Jinping Deep Underground Frontier Science and Dark Matter Key Laboratory of Sichuan Province Liangshan 615000China Yalong River Hydropower Development Company Ltd.288 Shuanglin RoadChengdu 610051China School of Physics Sun Yat-Sen UniversityGuangzhou 510275China Sino-French Institute of Nuclear Engineering and Technology Sun Yat-Sen UniversityZhuhai 519082China Department of Physics Yantai UniversityYantai 264005China Key Laboratory of Nuclear Physics and Ion-beam Application(MOE) Institute of Modern PhysicsFudan UniversityShanghai 200433China School of Physics Beihang UniversityBeijing 102206China Beijing Key Laboratory of Advanced Nuclear Materials and Physics Beihang UniversityBeijing 102206China Peng Huanwu Collaborative Center for Research and Education Beihang UniversityBeijing 100191China Southern Center for Nuclear-Science Theory(SCNT) Institute of Modern PhysicsChinese Academy of SciencesHuizhou 516000China School of Mechanical Engineering Shanghai Jiao Tong UniversityShanghai 200240China Department of Physics University of MarylandCollege ParkMaryland 20742USA State Key Laboratory of Particle Detection and Electronics University of Science and Technology of ChinaHefei 230026China Department of Modern Physics University of Science and Technology of ChinaHefei 230026China School of Physics Peking UniversityBeijing 100871Ch

The proton-proton(pp)fusion chain dominates the neutrino production in the *** uncertainty of the predicted pp neutrino flux is at the sub-percent level,whereas that of the best measurement is O(10%).In this study,for the first time,we measure solar pp neutrinos in the electron recoil energy range from 24 to 144 keV using the PandaX-4T commissioning data with 0.63 tonne×year *** pp neutrino flux is determined as(8.0±3.9(stat)±10.0(syst))×1010 s^(-1)cm^(-2),which is consistent with the Standard Solar Model and existing measurements,corresponding to an upper flux limit of 23.3×10^(10)s^(-1)cm^(-2)at 90%C.L..

关键词： solar pp neutrino flux PandaX-4T electron recoil spectrum

Class-Imbalanced Semi-Supervised Learning for Large-Scale Point Cloud Semantic Segmentation Via Decoupling Optimization

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Li, Mengtian Lin, Shaohui Wang, Zihan Shen, Yunhang Zhang, Baochang Ma, Lizhuang Visual Media Intelligent Processing Research Group Department of Film and Television Engineering Shanghai University Shanghai200072 China School of Computer Science and Technology East China Normal University Shanghai200062 China Key Laboratory of Advanced Theory and Application in Statistics and Data Science Ministry of Education China School of Automation Science and Electrical Engineering Beihang University Beijing100191 China Youtu Lab Tencent China

Semi-supervised learning (SSL), thanks to the significant reduction of data annotation costs, has been an active research topic for large-scale 3D scene understanding. However, the existing SSL-based methods suffer from severe training bias, mainly due to class imbalance and long-tail distributions of the point cloud data. As a result, they lead to a biased prediction for the tail class segmentation. In this paper, we introduce a new decoupling optimization framework, which disentangles feature representation learning and classifier in an alternative optimization manner to shift the bias decision boundary effectively. In particular, we first employ two-round pseudo-label generation to select unlabeled points across head-to-tail classes. We further introduce multi-class imbalanced focus loss to adaptively pay more attention to feature learning across head-to-tail classes. We fix the backbone parameters after feature learning and retrain the classifier using ground-truth points to update its parameters. Ex-✩Fully documented templates are available in the elsarticle package on CTAN. © 2023, The Authors. All rights reserved.

关键词： Semantic Segmentation

Filter Pruning for Efficient CNNs via Knowledge-driven Differential Filter Sampler

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lin, Shaohui Huang, Wenxuan Xie, Jiao Zhang, Baochang Shen, Yunhang Yu, Zhou Han, Jungong Doermann, David The School of Computer Science and Technology East China Normal University Shanghai China The Key Laboratory of Advanced Theory and Application in Statistics and Data Science Ministry of Education China The Department of Automation School of Aerospace Engineering Xiamen University Xiamen China The Institute of Artificial Intelligence Beihang University Beijing China Zhongguancun Laboratory Beijing China The Youtu Lab Tencent Shanghai China The School of Statistics East China Normal University Shanghai China The Department of Computer Science University of Sheffield United Kingdom The University at Buffalo BuffaloNY United States

Filter pruning simultaneously accelerates the computation and reduces the memory overhead of CNNs, which can be effectively applied to edge devices and cloud services. In this paper, we propose a novel Knowledge-driven Differential Filter Sampler (KDFS) with Masked Filter Modeling (MFM) framework for filter pruning, which globally prunes the redundant filters based on the prior knowledge of a pre-trained model in a differential and non-alternative optimization. Specifically, we design a differential sampler with learnable sampling parameters to build a binary mask vector for each layer, determining whether the corresponding filters are redundant. To learn the mask, we introduce masked filter modeling to construct PCA-like knowledge by aligning the intermediate features from the pre-trained teacher model and the outputs of the student decoder taking sampling features as the input. The mask and sampler are directly optimized by the Gumbel-Softmax Straight-Through Gradient Estimator in an end-to-end manner in combination with global pruning constraint, MFM reconstruction error, and dark knowledge. Extensive experiments demonstrate the proposed KDFS’s effectiveness in compressing the base models on various datasets. For instance, the pruned ResNet-50 on ImageNet achieves 55.36% computation reduction, and 42.86% parameter reduction, while only dropping 0.35% Top-1 accuracy, significantly outperforming the state-of-the-art methods. The code is available at https://***/Osilly/KDFS. Copyright © 2023, The Authors. All rights reserved.

关键词： Machine learning

Sparse generalized canonical correlation analysis: Distributed alternating iteration based approach

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Cai, Jia Lv, Kexin Huo, Junyi Huang, Xiaolin Yang, Jie School of Statistics and Mathematics Guangdong University of Finance & Economics Big Data and Educational Statistics Application Laboratory 21 Chisha Road Guangzhou Guangdong510320 China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University MOE Key Laboratory of System Control and Information Processing 800 Dongchuan Road Shanghai200240 China School of Electronics and Computer Science University of Southampton University Road SouthamptonSO17 1BJ United Kingdom

Sparse canonical correlation analysis (CCA) is a useful statistical tool to detect latent information with sparse structures. However, sparse CCA works only for two datasets, i.e., there are only two views or two distinct objects. To overcome this limitation, in this paper, we propose a sparse generalized canonical correlation analysis (GCCA), which could detect the latent relations of multiview data with sparse structures. Moreover, the introduced sparsity could be considered as Laplace prior on the canonical variates. Specifically, we convert the GCCA into a linear system of equations and impose 1 minimization penalty for sparsity pursuit. This results in a nonconvex problem on Stiefel manifold, which is difficult to solve. Motivated by Boyd’s consensus problem, an algorithm based on distributed alternating iteration approach is developed and theoretical consistency analysis is investigated elaborately under mild conditions. Experiments on several synthetic and real world datasets demonstrate the effectiveness of the proposed algorithm. Copyright © 2020, The Authors. All rights reserved.

关键词： Statistical mechanics

Generalized Estimating Equations for Hearing Loss data with Specified Correlation Structures

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wei, Zhuoran Zhu, Hanbing Curhan, Sharon Curhan, Gary Wang, Molin Department of Biostatistics Harvard T.H. Chan School of Public Health BostonMA United States Key Laboratory of Advanced Theory and Application in Statistics and Data Science-MOE School of Statistics East China Normal University Shanghai China Harvard Medical School BostonMA United States Channing Division of Network Medicine Department of Medicine Brigham and Women’s Hospital BostonMA United States Division of Renal Medicine Department of Medicine Brigham and Women’s Hospital BostonMA United States Department of Epidemiology Harvard T.H. Chan School of Public Health BostonMA United States

Due to the nature of pure-tone audiometry test, hearing loss data often has a complicated correlation structure. Generalized estimating equation (GEE) is commonly used to investigate the association between exposures and hearing loss, because it is robust to misspecification of the correlation matrix. However, this robustness typically entails a moderate loss of estimation efficiency in finite samples. This paper proposes to model the correlation coefficients and use second-order generalized estimating equations to estimate the correlation parameters. In simulation studies, we assessed the finite sample performance of our proposed method and compared it with other methods, such as GEE with independent, exchangeable and unstructured correlation structures. Our method achieves an efficiency gain which is larger for the coefficients of the covariates corresponding to the within-cluster variation (e.g., ear-level covariates) than the coefficients of cluster-level covariates. The efficiency gain is also more pronounced when the within-cluster correlations are moderate to strong, or when comparing to GEE with an unstructured correlation structure. As a real-world example, we applied the proposed method to data from the Audiology Assessment Arm of the Conservation of Hearing Study, and studied the association between a dietary adherence score and hearing loss. Copyright © 2023, The Authors. All rights reserved.

关键词： Audition