Search:
Match:
15 results
Research Paper#Statistics, Machine Learning, Estimation🔬 ResearchAnalyzed: Jan 3, 2026 06:12

Compound Estimation for Binomials

Published:Dec 31, 2025 18:38
1 min read
ArXiv

Analysis

This paper addresses the problem of estimating the mean of multiple binomial outcomes, a common challenge in various applications. It proposes a novel approach using a compound decision framework and approximate Stein's Unbiased Risk Estimator (SURE) to improve accuracy, especially when dealing with small sample sizes or mean parameters. The key contribution is working directly with binomials without Gaussian approximations, enabling better performance in scenarios where existing methods struggle. The paper's focus on practical applications and demonstration with real-world datasets makes it relevant.
Reference

“The paper develops an approximate Stein's Unbiased Risk Estimator (SURE) for the average mean squared error and establishes asymptotic optimality and regret bounds for a class of machine learning-assisted linear shrinkage estimators.”

PermalinkArXiv
Research Paper#Time Series Analysis, Matrix Factorization, Machine Learning🔬 ResearchAnalyzed: Jan 3, 2026 06:13

Modewise Additive Factor Model for Matrix Time Series

Published:Dec 31, 2025 18:24
1 min read
ArXiv

Analysis

This paper introduces a novel Modewise Additive Factor Model (MAFM) for matrix-valued time series, offering a more flexible approach than existing multiplicative factor models like Tucker and CP. The key innovation lies in its additive structure, allowing for separate modeling of row-specific and column-specific latent effects. The paper's contribution is significant because it provides a computationally efficient estimation procedure (MINE and COMPAS) and a data-driven inference framework, including convergence rates, asymptotic distributions, and consistent covariance estimators. The development of matrix Bernstein inequalities for quadratic forms of dependent matrix time series is a valuable technical contribution. The paper's focus on matrix time series analysis is relevant to various fields, including finance, signal processing, and recommendation systems.
Reference

“The key methodological innovation is that orthogonal complement projections completely eliminate cross-modal interference when estimating each loading space.”

PermalinkArXiv
Research Paper#Reinforcement Learning, Offline RL, Robustness, Sparsity🔬 ResearchAnalyzed: Jan 3, 2026 17:07

Sparse Offline RL Robust to Data Corruption

Published:Dec 31, 2025 10:28
1 min read
ArXiv

Analysis

This paper addresses the challenge of robust offline reinforcement learning in high-dimensional, sparse Markov Decision Processes (MDPs) where data is subject to corruption. It highlights the limitations of existing methods like LSVI when incorporating sparsity and proposes actor-critic methods with sparse robust estimators. The key contribution is providing the first non-vacuous guarantees in this challenging setting, demonstrating that learning near-optimal policies is still possible even with data corruption and specific coverage assumptions.
Reference

“The paper provides the first non-vacuous guarantees in high-dimensional sparse MDPs with single-policy concentrability coverage and corruption, showing that learning a near-optimal policy remains possible in regimes where traditional robust offline RL techniques may fail.”

PermalinkArXiv
Research Paper#Statistics, Clinical Trials, Bayesian Methods🔬 ResearchAnalyzed: Jan 3, 2026 09:28

Model-Assisted Bayesian Estimators for Ordinal Outcomes in RCTs

Published:Dec 30, 2025 19:53
1 min read
ArXiv

Analysis

This paper addresses the limitations of traditional methods (like proportional odds models) for analyzing ordinal outcomes in randomized controlled trials (RCTs). It proposes more transparent and interpretable summary measures (weighted geometric mean odds ratios, relative risks, and weighted mean risk differences) and develops efficient Bayesian estimators to calculate them. The use of Bayesian methods allows for covariate adjustment and marginalization, improving the accuracy and robustness of the analysis, especially when the proportional odds assumption is violated. The paper's focus on transparency and interpretability is crucial for clinical trials where understanding the impact of treatments is paramount.
Reference

“The paper proposes 'weighted geometric mean' odds ratios and relative risks, and 'weighted mean' risk differences as transparent summary measures for ordinal outcomes.”

PermalinkArXiv
Research Paper#Stochastic Differential Equations, Lévy Noise, Least Squares Estimation, Sparse Data🔬 ResearchAnalyzed: Jan 3, 2026 18:21

Least Squares Estimation for SDEs with Lévy Noise and Sparse Data

Published:Dec 30, 2025 05:58
1 min read
ArXiv

Analysis

This paper addresses a practical problem in financial modeling and other fields where data is often sparse and noisy. The focus on least squares estimation for SDEs perturbed by Lévy noise, particularly with sparse sample paths, is significant because it provides a method to estimate parameters when data availability is limited. The derivation of estimators and the establishment of convergence rates are important contributions. The application to a benchmark dataset and simulation study further validate the methodology.
Reference

“The paper derives least squares estimators for the drift, diffusion, and jump-diffusion coefficients and establishes their asymptotic rate of convergence.”

PermalinkArXiv
Research Paper#Model Checking, Concurrency, State Space Estimation🔬 ResearchAnalyzed: Jan 3, 2026 18:22

State Space Estimation for DPOR-based Model Checkers

Published:Dec 30, 2025 05:32
1 min read
ArXiv

Analysis

This paper addresses the challenging problem of estimating the size of the state space in concurrent program model checking, specifically focusing on the number of Mazurkiewicz trace-equivalence classes. This is crucial for predicting model checking runtime and understanding search space coverage. The paper's significance lies in providing a provably poly-time unbiased estimator, a significant advancement given the #P-hardness and inapproximability of the counting problem. The Monte Carlo approach, leveraging a DPOR algorithm and Knuth's estimator, offers a practical solution with controlled variance. The implementation and evaluation on shared-memory benchmarks demonstrate the estimator's effectiveness and stability.
Reference

“The paper provides the first provable poly-time unbiased estimators for counting traces, a problem of considerable importance when allocating model checking resources.”

PermalinkArXiv
Research Paper#Phylogenetics, Density Estimation, Machine Learning🔬 ResearchAnalyzed: Jan 3, 2026 18:49

Bandwidth Selection for Phylogenetic Tree Density Estimation

Published:Dec 29, 2025 13:01
1 min read
ArXiv

Analysis

This paper addresses the problem of bandwidth selection for kernel density estimation (KDE) applied to phylogenetic trees. It proposes a likelihood cross-validation (LCV) method for selecting the optimal bandwidth in a tropical KDE, a KDE variant using a specific distance metric for tree spaces. The paper's significance lies in providing a theoretically sound and computationally efficient method for density estimation on phylogenetic trees, which is crucial for analyzing evolutionary relationships. The use of LCV and the comparison with existing methods (nearest neighbors) are key contributions.
Reference

“The paper demonstrates that the LCV method provides a better-fit bandwidth parameter for tropical KDE, leading to improved accuracy and computational efficiency compared to nearest neighbor methods, as shown through simulations and empirical data analysis.”

PermalinkArXiv
Research Paper#Survival Analysis, Ranked Set Sampling, Statistical Methods🔬 ResearchAnalyzed: Jan 3, 2026 19:46

Ranked Set Sampling for Survival Analysis: A Unified Framework

Published:Dec 27, 2025 17:15
1 min read
ArXiv

Analysis

This paper addresses a significant gap in survival analysis by developing a comprehensive framework for using Ranked Set Sampling (RSS). RSS is a cost-effective sampling technique that can improve precision. The paper extends existing RSS methods, which were primarily limited to Kaplan-Meier estimation, to include a broader range of survival analysis tools like log-rank tests and mean survival time summaries. This is crucial because it allows researchers to leverage the benefits of RSS in more complex survival analysis scenarios, particularly when dealing with imperfect ranking and censoring. The development of variance estimators and the provision of practical implementation details further enhance the paper's impact.
Reference

“The paper formalizes Kaplan-Meier and Nelson-Aalen estimators for right-censored data under both perfect and concomitant-based imperfect ranking and establishes their large-sample properties.”

PermalinkArXiv
Research Paper#Fuzzy Logic, Statistics, Regression, Multicollinearity🔬 ResearchAnalyzed: Jan 3, 2026 19:57

Robust Estimation for Multicollinearity in Fuzzy Logistic Regression

Published:Dec 27, 2025 08:11
1 min read
ArXiv

Analysis

This paper tackles a common problem in statistical modeling (multicollinearity) within the context of fuzzy logic, a less common but increasingly relevant area. The use of fuzzy numbers for both the response variable and parameters adds a layer of complexity. The paper's significance lies in proposing and evaluating several Liu-type estimators to mitigate the instability caused by multicollinearity in this specific fuzzy logistic regression setting. The application to real-world fuzzy data (kidney failure) further validates the practical relevance of the research.
Reference

“FLLTPE and FLLTE demonstrated superior performance compared to other estimators.”

PermalinkArXiv
Research Paper#Reinforcement Learning, Large Language Models, KL Divergence, Regularization🔬 ResearchAnalyzed: Jan 3, 2026 23:59

KL Regularization in RL Training of LLMs: A Deep Dive

Published:Dec 26, 2025 04:20
1 min read
ArXiv

Analysis

This paper investigates the impact of different Kullback-Leibler (KL) divergence estimators used for regularization in Reinforcement Learning (RL) training of Large Language Models (LLMs). It highlights the importance of choosing unbiased gradient estimators to avoid training instabilities and improve performance on both in-domain and out-of-domain tasks. The study's focus on practical implementation details and empirical validation with multiple LLMs makes it valuable for practitioners.
Reference

“Using estimator configurations resulting in unbiased gradients leads to better performance on in-domain as well as out-of-domain tasks.”

PermalinkArXiv
Research#llm🔬 ResearchAnalyzed: Dec 25, 2025 04:07

Semiparametric KSD Test: Unifying Score and Distance-Based Approaches for Goodness-of-Fit Testing

Published:Dec 24, 2025 05:00
1 min read
ArXiv Stats ML

Analysis

This arXiv paper introduces a novel semiparametric kernelized Stein discrepancy (SKSD) test for goodness-of-fit. The core innovation lies in bridging the gap between score-based and distance-based GoF tests, reinterpreting classical distance-based methods as score-based constructions. The SKSD test offers computational efficiency and accommodates general nuisance-parameter estimators, addressing limitations of existing nonparametric score-based tests. The paper claims universal consistency and Pitman efficiency for the SKSD test, supported by a parametric bootstrap procedure. This research is significant because it provides a more versatile and efficient approach to assessing model adequacy, particularly for models with intractable likelihoods but tractable scores.
Reference

“Building on this insight, we propose a new nonparametric score-based GoF test through a special class of IPM induced by kernelized Stein's function class, called semiparametric kernelized Stein discrepancy (SKSD) test.”

PermalinkArXiv Stats ML
Research#llm🔬 ResearchAnalyzed: Dec 25, 2025 04:22

Generative Bayesian Hyperparameter Tuning

Published:Dec 24, 2025 05:00
1 min read
ArXiv Stats ML

Analysis

This paper introduces a novel generative approach to hyperparameter tuning, addressing the computational limitations of cross-validation and fully Bayesian methods. By combining optimization-based approximations to Bayesian posteriors with amortization techniques, the authors create a "generator look-up table" for estimators. This allows for rapid evaluation of hyperparameters and approximate Bayesian uncertainty quantification. The connection to weighted M-estimation and generative samplers further strengthens the theoretical foundation. The proposed method offers a promising solution for efficient hyperparameter tuning in machine learning, particularly in scenarios where computational resources are constrained. The approach's ability to handle both predictive tuning objectives and uncertainty quantification makes it a valuable contribution to the field.
Reference

“We develop a generative perspective on hyper-parameter tuning that combines two ideas: (i) optimization-based approximations to Bayesian posteriors via randomized, weighted objectives (weighted Bayesian bootstrap), and (ii) amortization of repeated optimization across many hyper-parameter settings by learning a transport map from hyper-parameters (including random weights) to the corresponding optimizer.”

PermalinkArXiv Stats ML
Research#Statistics🔬 ResearchAnalyzed: Jan 10, 2026 09:00

Debiased Inference for Fixed Effects Models in Complex Data

Published:Dec 21, 2025 10:35
1 min read
ArXiv

Analysis

This ArXiv paper explores methods for improving the accuracy of statistical inference in the context of panel and network data. The focus on debiasing fixed effects estimators is particularly relevant given their widespread use in various fields.
Reference

“The paper focuses on fixed effects estimators with three-dimensional panel and network data.”

PermalinkArXiv
Research#MLE🔬 ResearchAnalyzed: Jan 10, 2026 12:09

Analyzing Learning Curve Behavior in Maximum Likelihood Estimation

Published:Dec 11, 2025 02:12
1 min read
ArXiv

Analysis

This ArXiv paper investigates the learning behavior of Maximum Likelihood Estimators, a crucial aspect of statistical machine learning. Understanding learning curve monotonicity provides valuable insights into the performance and convergence properties of these estimators.
Reference

“The paper examines learning-curve monotonicity for Maximum Likelihood Estimators.”

PermalinkArXiv
Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 08:01

Point and interval estimators of a changepoint in stochastical dominance between two distributions

Published:Dec 9, 2025 17:14
1 min read
ArXiv

Analysis

This article, sourced from ArXiv, focuses on statistical methods for identifying and estimating change points in the stochastic dominance relationship between two probability distributions. The research likely explores the development and evaluation of point and interval estimators, which are crucial for understanding how the dominance relationship evolves over time or across different conditions. The use of 'stochastic dominance' suggests the study's relevance to fields where comparing distributions is essential, such as finance, economics, or risk management.

Key Takeaways

    Reference

    “”

    PermalinkArXiv