archive
Every paper Pith has read. Search by title, abstract, or pith.
2685 papers in stat.ML · page 16
-
The paper proves that any random variable whose moment generating function is bounded…
Sharp One-Dimensional Sub-Gaussian Comparison in Convex Order
-
Sufficient statistic reduces IB to low-dimensional equivalent
A Sufficient-Statistic Reduction of the Information Bottleneck to a Low-Dimensional Problem
-
Pruning cuts feature engineering time for high-dimensional data
SCOPE-FE: Structured Control of Operator and Pairwise Exploration for Feature Engineering
-
Laplace approximation restores Bayesian inference to tensor network kernels
Laplace Approximation for Bayesian Tensor Network Kernel Machines
-
ECG attributions mapped to 3D heart space raise localization accuracy
Validating the Clinical Utility of CineECG 3D Reconstructions through Cross-Modal Feature Attribution
-
Neural network classifier tops power in independence tests
Deep-testing: the case of dependence detection
-
Diffusion model assigns outlier probabilities to structural sensor readings
Probabilistic data quality assessment for structural monitoring data via outlier-resistant conditional diffusion model
-
Rejection sampling sets entropy curves to beat LLM RL saturation
Addressing Performance Saturation for LLM RL via Precise Entropy Curve Control
-
Rejection sampling enforces entropy curves to extend LLM RL gains
Addressing Performance Saturation for LLM RL via Precise Entropy Curve Control
-
The paper develops a co-learning framework that simultaneously learns a port-Hamiltonian…
Co-Learning Port-Hamiltonian Systems and Optimal Energy-Shaping Control
-
Alternating optimization yields passive stable controllers from data
Co-Learning Port-Hamiltonian Systems and Optimal Energy-Shaping Control
-
Online bandits allocate ad budgets from the first user
Budget-Constrained Causal Bandits: Bridging Uplift Modeling and Sequential Decision-Making
-
Adaptive CIs in two-groups model scale as σ(n^{-1/4} + ε term)
Adaptive Confidence Intervals in Efron's Gaussian Two-Groups Model
-
Representations from modeling and marginalizing environment variation achieve better…
Robust Representation Learning through Explicit Environment Modeling
-
Rule on glucose changes cuts updates in AI pancreas control
Application of Deep Reinforcement Learning to Event-Triggered Control for Networked Artificial Pancreas Systems
-
Rule-based glucose trigger cuts AP system communications
Application of Deep Reinforcement Learning to Event-Triggered Control for Networked Artificial Pancreas Systems
-
Revenue encoding segments customers for profitable product recommendations
Value-Aware Product Recommendation by Customer Segmentation using a suitable High-Dimensional Similarity Measure
-
Wasserstein post-processing removes subsampling bias in GLMM sampling
Safe, Scalable, and Accurate Bayes Posterior Sampling for Large-Data Generalized Linear Mixed Models
-
Teacher forcing inflates curvature in switching chaotic models
Teacher Forcing as Generalized Bayes: Optimization Geometry Mismatch in Switching Surrogates for Chaotic Dynamics
-
Imperfect rewards can help policy gradients avoid mediocre outputs
When Errors Can Be Beneficial: A Categorization of Imperfect Rewards for Policy Gradient
-
Small clean data purifies noisy labels for classification
Model-agnostic information transfer and fusion for classification with label noise
-
ELBO overfits regression when posterior covariance rank is restricted
Occam's Razor is Only as Sharp as Your ELBO
-
Domain generalization produces compact histopathology signatures
Magnification-Invariant Image Classification via Domain Generalization and Stable Sparse Embedding Signatures
-
Adaptive networks let one MCMC sampler handle many similar structural updates
Adaptive Meta-Learning Stochastic Gradient Hamiltonian Monte Carlo Simulation for Bayesian Updating of Structural Dynamic Models
-
Deflation-free scoring matches sequential accuracy on high-dim data
Deflation-Free Optimal Scoring
-
Residual spikes in PINNs locate regime transitions
Residual-loss Anomaly Analysis of Physics-Informed Neural Networks: An Inverse Method for Change-point Detection in Nonlinear Dynamical Systems with Regime Switching
-
Optimal Kelly wealth growth equals bipolar KL limit
The optimal betting wealth growth rate
-
Graph bandits tame regret via small effective dimension
Spectral bandits
-
Bandit regret scales as 1/sqrt(r) without knowing r
Online learning with Erd\H{o}s-R\'enyi side-observation graphs
-
New estimator tightens regret for sleeping bandits with random availability
Online combinatorial optimization with stochastic decision sets and adversarial losses
-
VLM judges rank responses well but cannot assign trustworthy scores
VLM Judges Can Rank but Cannot Score: Task-Dependent Uncertainty in Multimodal Evaluation
-
NTK networks hit optimal rates for adversarial Sobolev regression
Adversarial Robustness of NTK Neural Networks
-
Tail allocation recovers shortest single-interval conformal predictors
Tail allocation for conformal prediction intervals
-
Elite SVMs steer slacks toward trusted reference benchmarks
Elite-Driven Support Vector Machines for Classification
-
Ensemble smoother infers causality from future observations
A Continuous-Time Ensemble Kalman-Bucy Smoother for Causal Inference and Model Discovery
-
Latent model adapts FSC to maxima-nominated samples
Fractionally Supervised Classification with Maxima Nominated Samples
-
Conformal sets predict future conflict sequences with coverage
Conflict Forecasting via Conformal Prediction for Markov Processes
-
Ghost gaps in VC learning need only analytic measurability
Null Measurability at the Symmetrization Interface in VC Learning
-
Thompson sampling matches scalar bounds using only pairwise comparisons
A Finite Time Analysis of Thompson Sampling for Bayesian Optimization with Preferential Feedback
-
CoreFlow improves matrix generation from few samples via low-rank cores
CoreFlow: Low-Rank Matrix Generative Models
-
Meta-theorem reduces identifiability proofs to set intersections
A Unifying Framework for Unsupervised Concept Extraction
-
Evolutionary trees recover LLM training histories from weights
Analysis and Explainability of LLMs Via Evolutionary Methods
-
Recipe maps ReLU approximations to softmax attention bounds
Transformer Approximations from ReLUs
-
Multiclass sample complexity now tight with DS dimension
The Optimal Sample Complexity of Multiclass and List Learning
-
Active learning uses constant CoT per thinker with log-log thinkers
Learning to Think from Multiple Thinkers
-
IV inference valid without exact structural equations
Instrumental Variable Analysis Without Structural Equations
-
Density-first model yields MD trajectories with IR spectra
Enhancing molecular dynamics with equivariant machine-learned densities
-
Implicit exploration yields optimal regret without knowing the observation graph
Efficient learning by implicit exploration in bandit problems with side observations
-
-
StoSOO optimizes noisy functions without smoothness knowledge
Stochastic simultaneous optimistic optimization