Article
Computer Science, Artificial Intelligence
David S. Watson, Marvin N. Wright
Summary: The proposed method introduces a new estimator called Conditional Predictive Impact (CPI) for measuring the association between features and outcomes under reduced feature sets. Through various algorithm tests and simulations, it has been demonstrated that CPI performs favorably compared to alternative methods.
Article
Automation & Control Systems
Zhanrui Cai, Runze Li, Yaowu Zhang
Summary: This paper introduces a new index to measure conditional dependence which is distribution free, ranging from zero to one, robust to outliers, and has low computational cost. It is applicable for multivariate random vectors and discrete data, providing a useful statistical inference tool for various data.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Statistics & Probability
Torkel Erhardsson
Summary: This paper clarifies and refines the definition of a reciprocal random field on an undirected graph, introducing four new properties (factorizing, global, local, and pairwise reciprocal properties) and their relationships. It shows that these properties reduce to the well-known Markov properties for specific cases and derives conditions for each reciprocal property to imply the next stronger property. Furthermore, it demonstrates that the subgraph induced by the remaining nodes preserves all four properties with respect to the node set delta \ delta(0).
JOURNAL OF APPLIED PROBABILITY
(2023)
Review
Physics, Multidisciplinary
Jan Mielniczuk
Summary: This paper reviews the information theoretic tools and their application in feature selection, focusing on classification problems with discrete features. The authors discuss various ways of constructing counterparts to conditional mutual information and their properties and limitations. They propose a unified method based on truncation for the Mobius expansion of conditional mutual information. The paper also discusses the main approaches to feature selection using the introduced measures of conditional dependence, along with methods for assessing the quality of the obtained predictors, including recent results on asymptotic distributions of empirical criteria and advances in resampling.
Article
Multidisciplinary Sciences
Prathitha Kar, Sriram Tiruvadi-Krishnan, Jaana Mannik, Jaan Mannik, Ariel Amir
Summary: How cells regulate their cell cycles is a central question for cell biology. This paper presents a study on cell size regulation in Escherichia coli, using conditional independence tests and data on cell size at key cell cycle events. The results suggest that the division event is controlled by the onset of constriction at midcell, and that replication-related processes and additional cues beyond DNA replication play roles in this control.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2023)
Article
Environmental Sciences
German A. Villalba, Xu Liang, Yao Liang
Summary: The paper introduces a novel method to systematically select multiple reference gauges for daily streamflow time series estimations, addressing a fundamental challenge in incomplete record sites.
WATER RESOURCES RESEARCH
(2021)
Article
Economics
Hakon Otneim, Dag Tjostheim
Summary: This article introduces a new measure of conditional dependence called the local Gaussian partial correlation (LGPC). Compared to traditional partial correlation coefficients, LGPC can better describe conditional dependence in a wide range of populations and has some useful and novel properties. LGPC can also be used to study departures from conditional independence in specific parts of the distribution.
JOURNAL OF BUSINESS & ECONOMIC STATISTICS
(2022)
Article
Mathematics
Peter J. Hammond, Yeneng Sun
Summary: This study explores the conditions under which a process defined by a continuum of random variables with non-degenerate idiosyncratic risk is jointly measurable, and provides a specific sigma algebra condition. Applications of the research include new characterizations and results related to conditional independence.
ADVANCES IN MATHEMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Kun Kuang, Haotian Wang, Yue Liu, Ruoxuan Xiong, Runze Wu, Weiming Lu, Yueting Zhuang, Fei Wu, Peng Cui, Bo Li
Summary: This paper addresses the problem of stable prediction across unknown test data, where the test distribution might be different from the training one. An algorithm based on conditional independence tests is proposed to screen out non-causal features and reduce spurious correlations by leveraging a seed variable, increasing the stability of prediction across unknown test data. Experimental results demonstrate that the proposed algorithm outperforms state-of-the-art methods for stable prediction across unknown test data.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Statistics & Probability
Matey Neykov, Sivaraman Balakrishnan, Larry Wasserman
Summary: The study focuses on the problem of conditional independence testing of X and Y given Z, considering smoothness assumptions on conditional distributions and testing difficulty. Lower and upper bounds were derived on the critical radius of separation between null and alternative hypotheses in the total variation metric.
ANNALS OF STATISTICS
(2021)
Article
History & Philosophy Of Science
Gregory Wheeler, Fabio G. Cozman
Summary: The purpose of this paper is to highlight that when conditional probabilities are adopted as the primitive concept of probability, there may be imprecise probability values and questions that fail to have numerically precise answers, even in very ordinary circumstances.
Article
Statistics & Probability
Ilmun Kim, Matey Neykov, Sivaraman Balakrishnan, Larry Wasserman
Summary: In this paper, the authors investigate the theoretical foundations of local permutation tests for testing conditional independence and specifically focus on binning-based statistics. They establish conditions for the universal validity of the local permutation method and introduce a double-binning permutation strategy to improve the effectiveness of the test. Simulation results are presented to support their theoretical findings.
ANNALS OF STATISTICS
(2022)
Article
Physics, Multidisciplinary
Lei Zan, Anouar Meynaoui, Charles K. Assaad, Emilie Devijver, Eric Gaussier
Summary: In this study, we focus on mixed data and propose a novel method, CMIh, to estimate conditional mutual information. We also introduce a new local permutation test, LocAT, which is well suited for mixed data. Our experiments demonstrate the good performance of CMIh and LocAT in accurately estimating conditional mutual information and detecting conditional (in)dependence for mixed data.
Article
Statistics & Probability
Jun Tao, Bing Li, Lingzhou Xue
Summary: We introduce a nonparametric graphical model for discrete node variables based on additive conditional independence, develop an estimator and establish its consistency, and uncover a deeper relation between additive conditional independence and conditional independence. The new method is evaluated through simulation experiments and analysis of a real dataset.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Biology
Tamas Spisak
Summary: The proposed partial confounder test provides a strict control for type I errors and high statistical power, even in the presence of nonnormally and nonlinearly dependent predictions. It can reveal previously unreported confounders and identify cases where state-of-the-art confound mitigation approaches fail.
Article
Engineering, Industrial
Yifu Li, Hongyue Sun, Xinwei Deng, Chuck Zhang, Hsu-Pin (Ben) Wang, Ran Jin
Article
Statistics & Probability
Sumin Shen, Lulu Kang, Xinwei Deng
Article
Computer Science, Interdisciplinary Applications
Xiaoning Kang, Xinwei Deng
JOURNAL OF STATISTICAL COMPUTATION AND SIMULATION
(2020)
Article
Statistics & Probability
Xiaoning Kang, Xinwei Deng, Kam-Wah Tsui, Mohsen Pourahmadi
INTERNATIONAL STATISTICAL REVIEW
(2020)
Article
Statistics & Probability
Qian Wu, Xinwei Deng, Shiren Wang, Li Zeng
Summary: This article introduces a constrained varying-coefficient modeling method for the fabrication of artificial soft tissues to obtain novel biomaterials with desired properties by adjusting process parameters, incorporating expert knowledge in model estimation. The proposed model has a semiparametric structure and includes constraints on model coefficients based on expert knowledge.
Article
Statistics & Probability
Shuyu Chu, Huijing Jiang, Zhengliang Xue, Xinwei Deng
Summary: In the pricing of customized products, accurately predicting the purchase behavior of different customers for personalized requests is challenging, requiring the construction of distinct models for data analysis. An adaptive convex clustering method is proposed to segment data and fit models simultaneously, ensuring data points with similar model structures are grouped together.
Review
Computer Science, Artificial Intelligence
Xiaoning Kang, Xinwei Deng
WILEY INTERDISCIPLINARY REVIEWS-DATA MINING AND KNOWLEDGE DISCOVERY
(2020)
Article
Engineering, Industrial
Hao Wang, Qiong Zhang, Kaibo Wang, Xinwei Deng
QUALITY ENGINEERING
(2020)
Article
Engineering, Industrial
Xiaoning Kang, Xiaoyu Chen, Ran Jin, Hao Wu, Xinwei Deng
Article
Statistics & Probability
Xiaoning Kang, Xinwei Deng
Summary: The article explores methods for estimating large sparse covariance matrices and proposes a new approach to address the issue of variable order, ensuring positive definiteness of the estimator while capturing the sparse structure of the covariance matrix. The merits of the method are demonstrated through simulation studies and a practical example.
CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE
(2021)
Article
Statistics & Probability
Yiou Li, Xinwei Deng
Summary: Generalized linear models (GLMs) are widely used in statistical analysis, and studying optimal designs for improving prediction accuracy is crucial. This work proposes Elastic I-optimality as a prediction-oriented design criterion for GLMs, develops an efficient algorithm, and conducts numerical examples to evaluate feasibility and computational efficiency.
CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE
(2021)
Article
Ergonomics
Huiying Mao, Xinwei Deng, Honggang Jiang, Liang Shi, Hao Li, Liheng Tuo, Donghai Shi, Feng Guo
Summary: This study evaluated crash risk factors for ride-hailing drivers and found that crash history, the percentage of long-shift bookings, driving distance, operations during peak hours, years of being a ride-hailing driver, and passenger rating were significantly associated with crash risk. The findings provide critical information for the development of safety countermeasures, driver education programs, and safety regulations for the ride-hailing industry.
ACCIDENT ANALYSIS AND PREVENTION
(2021)
Article
Mathematics, Applied
Weijun Xie, Xinwei Deng
SIAM JOURNAL ON OPTIMIZATION
(2020)
Article
Computer Science, Information Systems
Vanessa Cedeno-Mieles, Zhihao Hu, Yihui Ren, Xinwei Deng, Abhijin Adiga, Christopher Barrett, Noshir Contractor, Saliya Ekanayake, Joshua M. Epstein, Brian J. Goode, Gizem Korkmaz, Chris J. Kuhlman, Dustin Machi, Michael W. Macy, Madhav V. Marathe, Naren Ramakrishnan, S. S. Ravi, Parang Saraf, Nathan Self
SOCIAL NETWORK ANALYSIS AND MINING
(2020)
Article
Statistics & Probability
Sumin Shen, Zhiyang Zhang, Xinwei Deng
JOURNAL OF STATISTICAL THEORY AND PRACTICE
(2020)