archive
Every paper Pith has read. Search by title, abstract, or pith.
2684 papers in stat.ML · page 5
-
NeuroMAS models multi-agent LLMs as trainable neural networks
NeuroMAS: Multi-Agent Systems as Neural Networks with Joint Reinforcement Learning
-
β-TCVAE recovers nonlinear brain networks from fMRI
Isolating Nonlinear Independent Sources in fMRI with $\beta$-TCVAE Models
-
SaaS with caps needs actuarial pricing for tail risks
Your SaaS Is an Insurance Product: A Modeling Framework
-
Hypothesis test picks samples to unlearn data domains
Statistical Unlearning of Distributions: A Hypothesis Testing Approach
-
Score matching closes nonlinear filter moments without integrals
The Score Kalman Filter
-
Weight decay slows sharpening and triggers architecture phase shifts
Does Weight Decay Enhance Training Stability?
-
Reallocating HCV treatments cuts costs by millions for HIV patients
Policy Learning with Observational Data: The Case of Hepatitis C Treatment for HIV/HCV Co-Infected Patients
-
Isotonic regression calibrates Deep Cox survival predictions
Isotonic Survival Regression: Calibrated Survival Distributions from Deep Cox Models
-
Cubing strategy finds stable regions for calibrated dropout uncertainty
A Cubing Strategy for Identifying Stable Hyperparameter Regions for Uncertainty Quantification in Spatial Deep Learning
-
Stein operator learns divergences without Jacobians in diffusion models
StAD: Stein Amortized Divergence for Fast Likelihoods with Diffusion and Flow
-
FRESH re-calibrates patient models to match population aggregates geometrically
FRESH: Information-Geometric Calibration of Patient-Level Models to Aggregate Evidence
-
Dependent observations reach zero entropy after O(log(1/Pmin)) samples
Breaking the Finite-Sample Barrier in Entropy Coupling
-
Privacy reduces CVaR effective samples to ε n τ
The Privacy Price of Tail-Risk Learning: Effective Tail Sample Size in Differentially Private CVaR Optimization
-
Quadrature lets deep networks model continuous survival hazards scalably
A Scalable Nonparametric Continuous-Time Survival Model through Numerical Quadrature
-
Skew-adaptive conformal prediction keeps validity with tilted intervals
Skew-adaptive conformal prediction
-
Surrogates err 10 times more at extremes in stochastic heat flow
A numerical study into neural network surrogate model performance for uncertainty propagation
-
The paper proposes a variational quantum classifier using amplitude encoding stabilized…
SAFE Quantum Machine Learning with Variational Quantum Classifiers
-
Exponential integrator yields dimension-uniform KL bounds for mixtures
Dimension-Uniform Discretization Analysis of Preconditioned Annealed Langevin Dynamics for Multimodal Gaussian Mixtures
-
XAI explanations fall short of grounds for overturning decisions
Explainable AI Isn't Enough! Rethinking Algorithmic Contestability
-
Tree properties testable with sub-quadratic covariance queries
Testing properties of trees in graphical models with covariance queries
-
Kalman filter on cell complexes recovers topology from partial data
Topological Kalman Filtering on Cell Complexes
-
Kalman filter on cell complexes tracks network states via boundary diffusion
Topological Kalman Filtering on Cell Complexes
-
Node-private algorithms recover communities in SBMs if epsilon grows fast enough
Node-private community estimation in stochastic block models: Tractable algorithms and lower bounds
-
Domain shifts traced to small feature sets via density anomalies
Unsupervised Domain Shift Detection with Interpretable Subspace Attribution
-
Non-log-concave sampling matches log-concave dimension dependence
Complexity of Non-Log-Concave Sampling in Fisher Information
-
Score-based models match intrinsic Wasserstein rates on manifolds
Intrinsic Wasserstein Rates for Score-Based Generative Models on Smooth Manifolds
-
Neural nets output conservative Gaussians that compose safely
Learning Context-conditioned Gaussian Overbounds for Convolution-Based Uncertainty Propagation
-
MVP algorithm tightens regret bounds for RL with changing action sets
Tighter Regret Bounds for Contextual Action-Set Reinforcement Learning
-
Smooth function fixes TCAV variance that refuses to decay
$\alpha$-TCAV: A Unified Framework for Testing with Concept Activation Vectors
-
Heterogeneity identifies DAG orderings up to two permutations
Leveraging heterogeneity for identifiability: Bayesian order-based learning of multiple DAGs
-
Risk-aware offline policies match 1/sqrt(n) optimality rates
Pessimistic Risk-Aware Policy Learning in Contextual Bandits
-
Log-memory sketch counts distinct noisy objects
MaxSketch: Robust Distinct Counting in Streams via Random Projections
-
Pretrained network delivers survival predictions without tuning
SurvivalPFN: Amortizing Survival Prediction via In-Context Bayesian Inference
-
TMLE borrows from full trial to sharpen subgroup estimates
Improving the Efficiency of Subgroup Analysis in Randomized Controlled Trials with TMLE
-
DiMS sampler exactly hits connected neural net minima
Don't Stop Me Yet: Sampling Loss Minima via Dissipative Riemannian Mechanics
-
Reasoning models take different paths
Reasoning Models Don't Just Think Longer, They Move Differently
-
Oracle price map learning yields sharp regret in semiparametric pricing
Harnessing Unimodality in Semiparametric Contextual Pricing via Oracle Price Map Learning
-
Convex potential targets population-level balance in MoE
$\phi$-Balancing for Mixture-of-Experts Training
-
JEPA auxiliaries alter LLM geometry without task gains
Representation Without Reward: A JEPA Audit for LLM Fine-Tuning
-
Local surrogate refinement matches full-model accuracy in 100D rare-event PDEs
Proposal-Guided Greedy Surrogate Refinement for PDE-Driven High-Dimensional Rare-Event Estimation
-
Augmentation strength steers representations along distinct geometric paths
How Data Augmentation Shapes Neural Representations
-
EHR tables sharpen timing in text-based clinical timelines
Text Knows What, Tables Know When: Clinical Timeline Reconstruction via Retrieval-Augmented Multimodal Alignment
-
RoSHAP gives stable rankings by summarizing SHAP distributions
RoSHAP: A Distributional Framework and Robust Metric for Stable Feature Attribution
-
Logging policies optimize off-policy estimates by balancing rewards and coverage
Logging Policy Design for Off-Policy Evaluation
-
Optimal logging policies minimize OPE error via reward-coverage balance
Logging Policy Design for Off-Policy Evaluation
-
Anomaly detection uncovers refinery LP errors and opportunities
From Data to Action: Accelerating Refinery Optimization with AI
-
AGOP from kernel regression recovers central subspace with fewer samples
Average Gradient Outer Product in kernel regression provably recovers the central subspace for multi-index models
-
The paper introduces ICGPS, which uses meta-trained generative models for in-context…
In-Context Learning for Data-Driven Censored Inventory Control
-
Min-Max-IRL reaches fast O(n^{-1}) rates without exploration
Fast Rates for Inverse Reinforcement Learning
-
Sequential feature recovery produces power-law scaling
Scaling Laws from Sequential Feature Recovery: A Solvable Hierarchical Model