archive
Every paper Pith has read. Search by title, abstract, or pith.
6897 papers in stat · page 5
-
Optimizer memory makes shuffle order first-order fine-tuning noise
Optimizer Memory Makes Shuffle Order a First-Order Source of Fine-Tuning Noise
-
Mixture models distinguish mild from gross anomalies in circular data
Modelling and detecting mild and gross anomalies in circular data via double-contaminated models
-
Gradient tweak keeps primary descent intact while meeting secondary goals
Not All Objectives Are Born Equal: Priority-Constrained Descent for Hierarchical Multi-Objective Optimization
-
Expert priors enter best-subsets MIO as log-odds penalties
A Mathematical Optimization Approach for Expert-Informed Bayesian Best Subset Selection
-
Bayesian model segments satellite images without local labels
Scalable Bayesian Spatial Mixture Modelling for Remote Sensing Image Segmentation
-
Unsupervised maps cut regional coverage gaps in conformal prediction
Self-Organized Conformal Prediction: Reducing Regional Coverage Gaps with Unsupervised Group Discovery
-
Bayesian CDD alone stays accurate on gene directions across all DREAM5 networks
Bayesian Copula Directional Dependence is Cross-Network Robust for Gene-Regulatory Pair Direction: A Benchmark Study on DREAM5
-
Data on 37 nurses disproves modeling assumption in roster analysis
Critique of "Use of roster charts in the investigation and prosecution of nurses ..." by John O' Quigley
-
Language model embeddings beat hand-crafted features in car insurance pricing
Semantic insurance pricing with large language models
-
Symbolic trees generalize at rate L^d over sqrt(n)
Sample Complexity of Scientific Discovery: PAC Learnability of Compositional Function Trees
-
Gradient boosting extended to vector outputs in tree leaves
Gradient boosting with vector-valued leafs
-
Attention operator compresses distributions losslessly for Transformers
Generalization Analysis of Transformers in Distribution Regression
-
Model tracks Sri Lankan vegetable price surges at 86 percent accuracy on unseen 2024 data
When Prices Double in a Week: Forecasting of Agricultural Volatility in Import-Isolated Markets
-
Gaussian VI preconditions HMC and VAE guides MH sampling
Using Variational Inference to Improve the Efficiency of MCMC Algorithms
-
Abstention budget turns polynomial error decay exponential in Bayesian arm ID
Bayesian Best-Arm Identification with Abstention: A Polynomial-to-Exponential Phase Transition
-
Gauge-equivariant preconditioner cuts LM loss gap to 0.67 from 5.88
Dead-Direction Conditioners: Gauge-Equivariant Preconditioning for Deep Networks
-
Multivariate BART obtains first contraction rates with joint residual dependence
Multivariate Varying-Coefficient BART with Graphical Horseshoe Priors
-
Hutchinson-free objective cuts variance in few-step flow models
Few-Step Boltzmann Generators via Scalable Likelihood Flow Maps
-
Mixture posteriors adapt to true K with extra mass vanishing at n^{-1/2}
Posterior concentration and adaptation of the mixing measure in Dirichlet process mixtures
-
Panel flow matching pools sparse longitudinal data into full distributions
Panel Flow Matching: A Generative Approach to Learning Distributions of Longitudinal Data
-
spca R package produces uncorrelated sparse principal components
spca: An R package to Compute Least Squares Sparse Principal Components
-
Relaxed noise assumptions yield directed brain connectivity estimates
Connectivity Estimation using Stochastic Graph Heat Modelling
-
This paper develops a machine learning estimator for heterogeneous causal effects within…
Learning heterogeneous treatment effects under principal stratification
-
Model gives closed forms for discrete circular plus linear data
On Modeling Cylindrical Data with a Discrete Circular Component and Its Environmental Applications
-
Beta-tree test flags local model deviations with finite-sample bounds
Beta-trees for testing multivariate goodness-of-fit and localizing deviations from a model
-
Regret statistic classifies algo liquidity demand from data alone
Liquidity-Based Audit of Algorithmic Trading Strategies
-
OLS makes three recursive causal estimators identical in finite samples
Generated outcomes as generated regressors: Equivalences in recursive causal estimation
-
sBDCA solves LTS up to 3.25x faster than Fast-LTS with better objectives
Faster than Fast-LTS: Robust Regression and Outlier Detection with DC Programming
-
Specialized clinical AI beats general models by 25-39 points on real questions
Expert Evaluation of Clinical AI Tools on Real Point-of-Care Clinical Queries
-
Latent GP calibration achieves 95% coverage in aerodynamic uncertainty
A Bayesian latent Gaussian process framework for aerodynamic uncertainty quantification
-
Factor indeterminacy resolves at infinite feature scale
Perspectives on Latent Factor Indeterminacy and its Implications for Data Representation
-
Selectivity correction shrinks effects to 12-21% of published averages
Literature Review and Evidence Aggregation: a Toolkit for Applied Micro
-
Methods correct errors in both outcomes and covariates
Methods to address measurement error in both Outcome and Covariates
-
Anti-symmetric perturbation cuts fluctuation constant in non-reversible Langevin MC
Variance Reduction for Stochastic Gradient Generalized Non-reversible Langevin Monte Carlo Algorithms
-
LLM probe unifies EHR modalities for 87.69% ICD accuracy
Primary ICD Category Prediction using LLM-based Probing
-
Known sampling designs yield unbiased ML predictions without models
On design-unbiased algorithmic Machine Learning
-
Imputation ensembles form stable connected structures not random clouds
Topological reconstruction of Rubin multiple imputation via coarse proximity, Seifert van Kampen gluing and Hurewicz invariants
-
Measurement error in latent traits biases treatment effect estimates
Measurement Induced Confounding
-
Latent trait measurement error biases causal estimates
Measurement Induced Confounding
-
Estimator recovers causal effects despite confounding and missing data
Inferring Comprehensive Cohort Causal Effects in the Presence of Unmeasured Confounding and Missing Outcomes
-
Ray-based model scales projected Gaussians to high-dimensional compositional data
Composition as Direction: An Active-Set Ray-Based Model for Sparse High-Dimensional Compositional Data
-
Common stochastic fix fails for full conformal prediction
Full Conformal Prediction under Stochastic Non-Conformity Measure
-
IPW reframed as KL reweighting for post-Bayesian bias correction
Inverse Probability Weighting in a Post-Bayesian World
-
Nine LLM families show 90.3% consistent virtue rankings
Aristotelian Virtue Profiling of LLMs through Ethical Dilemmas
-
Stein operator derived for symmetric matrix normal from OU generator
Stein's method for the symmetric matrix normal distribution with an application to the approximation of the Wishart law
-
Extra samples stop helping answer selection after a few dozen draws
When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling
-
Adaptive hard thresholding attains logarithmic regret for online quantile regression
Adaptive Iterative Hard Thresholding for Online High-dimensional Quantile Regression
-
Shape regularity required for optimal local regression rates
Revisiting local regression: shape regularity, uniform rates, and the limits of random splits
-
Mamba model improves EHR patient subtyping over baselines
Improving Patient Subtyping on Longitudinal Data using Representations from Mamba-based Architecture
-
Bootstrap method enables non-asymptotic prediction-powered inference
A bootstrap approach to prediction-powered inference