Methodology creates holdout and test/train sets for ML studies, preserving data for future research.
problem Preserving data for future research studies that are analysis-naive.
method Modification of k-fold cross-validation, randomization, and three-way split (holdout, test, training).
result Efficiently creates holdout and test/train sets without forcing.
Holdout set improves risk score accuracy without biasing predictions.
problem Updating risk scores can lead to biased estimates when directly applied.
method Use a holdout set of non-intervention population to update risk scores.
result Optimal holdout size reduces adverse outcomes to optimal level.
AI agents improve forecast combination but require transparency.
problem AI coding agents increase flexibility in empirical economics, leading to hidden degrees of freedom.
method Adapted open-source agent-loop architecture to empirical economics workflow, adding post-search holdout evaluation.
result Multiple agent runs outperform standard benchmarks in rolling evaluation but not all on post-search holdout.
Paper analyzes holdout cross-validation for large non-Gaussian covariance estimation.
problem Estimating large covariance matrices for non-Gaussian data.
method Use of Weingarten calculus and Ledoit-Péché formula for theoretical error derivation.
result Optimal train-test split ratio is proportional to square root of matrix dimension.
In statistical learning, a dataset is often partitioned into two parts: the training set and the holdout (i.e., testing) set. For instance, the training set is used to learn a predictor, and then the holdout set is used for estimating the accuracy of the predictor on the true distribution. However, often in practice, t…
AI agents improve forecast combination in empirical economics.
problem Hidden researcher degrees of freedom in AI-generated code.
method Adapted agent-loop architecture to empirical economics, added holdout evaluation.
result Independent agent searches find better forecast methods than benchmarks.
The study introduces a holdout-based framework to assess synthetic data fidelity and privacy.
problem Evaluating the quality and privacy of synthetic data solutions for mixed-type tabular data.
method Holdout-based empirical assessment framework measuring fidelity and privacy risk.
result Synthetic data samples are as close to the training as to the holdout data, indicating generalization and independence from individual records.
We compare and contrast two approaches to validating a trained classifier while using all in-sample data for training. One is simultaneous validation over an organized set of hypotheses (SVOOSH), the well-known method that began with VC theory. The other is withhold and gap (WAG). WAG withholds a validation set, trains…
Optimal data splitting improves covariance matrix estimation in large datasets.
problem Improving large covariance matrix estimation in high-dimensional settings.
method Focus on holdout method, derive closed-form error expression, connect to eigenvalue variance.
result Optimal train-test split scales as square root of matrix dimension.
The study compares parametric and nonparametric models for estimating mean-variance mixtures and finds that nonparametric models perform better.
problem Estimating the distribution of a normal mean-variance mixture under uncertainty.
method Comparison of six parametric mixing laws with a grid nonparametric maximum likelihood estimator, using a paired block bootstrap for score comparison.
result Nonparametric models outperform parametric models in estimating the distribution of a normal mean-variance mixture.
Excessive reuse of holdout data can lead to overfitting. However, there is little concrete evidence of significant overfitting due to holdout reuse in popular multiclass benchmarks today. Known results show that, in the worst-case, revealing the accuracy of k adaptively chosen classifiers on a data set of size n al…
Memory is an important aspect of intelligence and plays a role in many deep reinforcement learning models. However, little progress has been made in understanding when specific memory systems help more than others and how well they generalize. The field also has yet to see a prevalent consistent and rigorous approach f…
Drug-drug interactions are preventable causes of medical injuries and often result in doctor and emergency room visits. Computational techniques can be used to predict potential drug-drug interactions. We approach the drug-drug interaction prediction problem as a link prediction problem and present two novel methods fo…
Jointly tuning ensemble models improves performance and uncertainty calibration.
problem Improving both predictive performance and uncertainty calibration in deep ensembles.
method Investigated the impact of jointly tuning weight decay, temperature scaling, and early stopping.
result Jointly tuning ensemble models generally matches or improves performance, with significant variation across tasks.
Reshuffling splits improves hyperparameter optimization's generalization performance.
problem Improving peak performance of machine learning models through better hyperparameter optimization.
method Reshuffling splits for every hyperparameter configuration improves generalization performance.
result Reshuffling leads to better generalization performance compared to fixed splits.
Paper examines fairness of data augmentation methods, finding vanilla Mixup outperforms Fair Mixup.
problem Improving fairness in classification models with limited minority data.
method Uses multicalibration to rigorously evaluate and improve data augmentation methods for classification fairness.
result Vanilla Mixup outperforms Fair Mixup and baseline methods in fairness and accuracy, especially with small minority groups.
Recent studies have shown that tuning prediction models increases prediction accuracy and that Random Forest can be used to construct prediction intervals. However, to our best knowledge, no study has investigated the need to, and the manner in which one can, tune Random Forest for optimizing prediction intervals { thi…
New method uses unlabeled data to improve generalization bounds for deep learning.
problem Vacuous guarantees and shrinking holdout sets for overparameterized models.
method Augmenting labeled training set with unlabeled data and training as usual.
result Proves tight upper bounds on true risk for 0-1 empirical risk minimization.
Cross-validation pitfalls in change-point regression are addressed with new approaches.
problem Cross-validation's prediction error-based criterion may lead to under- or over-estimation of change-points.
method Proposes two approaches: absolute error loss and modified holdout sets.
result Consistent estimation of the number of change-points under certain conditions.
Analyzes error sources in global feature effect estimation methods.
problem Unexplored error sources in global feature effect estimation methods.
method Systematic, estimator-level analysis of bias and variance.
result Holdout data is theoretically cleanest, but estimation variance depends on sample size and model characteristics.
In this work, we develop a novel regularizer to improve the learning of long-range dependency of sequence data. Applied on language modelling, our regularizer expresses the inductive bias that sequence variables should have high mutual information even though the model might not see abundant observations for complex lo…
We present a new modeling technique for solving the problem of ecological inference, in which individual-level associations are inferred from labeled data available only at the aggregate level. We model aggregate count data as arising from the Poisson binomial, the distribution of the sum of independent but not identic…
Replicated and validated Rank-N-Contrast for robust regression.
problem Deep regression models struggle with continuous sample orders.
method Contrastive learning of continuous representations by ranking samples.
result Improved performance and robustness of RNC framework.
PAC-Bayesian method improves deep learning generalization in medical imaging.
problem Overfitting in deep networks for medical imaging datasets.
method PAC-Bayesian framework applied to large (stochastic) networks.
result PAC-Bayesian bounds are competitive and more explainable than simpler methods.
This paper optimizes perpetual contract liquidity by accounting for funding rates.
problem Optimal liquidity provision for perpetual contracts with stochastic funding rates.
method Formulated a control problem, solved with a HJB scheme, and calibrated on real data.
result Funding-aware market making improves performance and reduces inventory risk.
Medical research is risky and expensive. Drug discovery, as an example, requires that researchers efficiently winnow thousands of potential targets to a small candidate set for more thorough evaluation. However, research groups spend significant time and money to perform the experiments necessary to determine this cand…
We propose the Sobolev Independence Criterion (SIC), an interpretable dependency measure between a high dimensional random variable X and a response variable Y . SIC decomposes to the sum of feature importance scores and hence can be used for nonlinear feature selection. SIC can be seen as a gradient regularized Integr…
Study evaluates how well question-answering models generalize to new data types.
problem Generalization of question-answering models to new data types.
method Constructed new test sets from different domains and evaluated models' performance.
result Models show significant performance drops when tested on new data types.
A framework for private prediction sets using conformal prediction and differential privacy.
problem Jointly addressing reliability and privacy in machine learning predictions.
method Split conformal prediction with privatized quantile subroutine.
result Private prediction sets can be generated from privately-trained models.
Study evaluates neural networks for corporate credit rating assessment.
problem Improving machine learning algorithms for credit assessment.
method Analysis of four neural network architectures (MLP, CNN, CNN2D, LSTM) on financial data from energy, financial, and healthcare sectors.
result LSTM architecture consistently outperforms others in predicting corporate credit ratings.
Malaria is a female anopheles mosquito-bite inflicted life-threatening disease which is considered endemic in many parts of the world. This article focuses on improving malaria detection from patches segmented from microscopic images of red blood cell smears by introducing a deep convolutional neural network. Compared …
Paper evaluates synthetic retail data for fidelity, utility, and privacy.
problem Ensuring accurate synthetic data in retail.
method Differentiates between continuous and discrete data, measures fidelity and utility, and uses Differential Privacy for privacy.
result Validated framework for reliable and scalable synthetic data evaluation.
Reducing volatility proxy improves apparent market correlation dynamics.
problem Attributing apparent slow collective market dynamics to intrinsic or driver inheritance.
method Coupled Ornstein-Uhlenbeck model with VIX proxy, decomposing and controlling for autocorrelation.
result VIX-coupled model reduces effective relaxation time from 298 to 61 trading days, improving fit over bare mean reversion.
Novel framework uses synthetic data to quantify uncertainty in complex data.
problem Uncertainty quantification in complex, unstructured data.
method Perturbation-Assisted Sample Synthesis (PASS) and Perturbation-Assisted Inference (PAI) framework.
result Statistically guaranteed validity in inference, enhancing reliability of synthetic data.
Deep neural networks predict prostate motion from MR images.
problem Predicting prostate motion during ultrasound-guided interventions.
method Biomechanically-trained deep neural networks on unstructured nodes.
result Trained networks yield near real-time inference with 0.017 mm error.
Interpretable additive models outperform complex DL and hybrid pipelines for air quality forecasting.
problem Accurate forecasting of urban air pollution for public health and policy guidance.
method Investigated lightweight additive models (FBP, NP) vs. deep learning and hybrid pipelines on Beijing PM2.5 and PM10 data.
result Facebook Prophet consistently outperformed NeuralProphet and traditional models, achieving high R2 values. The paper offers a method to create prediction sets with uncertainty control.
problem Calibrating and communicating uncertainty in machine learning predictions.
method Distribution-free, risk-controlling prediction sets using a holdout set to calibrate set sizes.
result Explicit finite-sample guarantees for error control in various machine learning tasks.
Tests for overfitting in machine learning models.
problem Overfitting in high complexity models.
method Hypothesis test using concentration bounds.
result Valid test for identifying overfitting.
New framework learns complex AI attitudes from heterogeneous data.
problem Heterogeneous ordinal structure in AI attitudes, poorly captured by existing methods.
method Monotone Gaussian score embedding, BNP complexity discovery, confirmatory fixed-K estimation.
result Reduced holdout MSE by 25.8% over single-graph baseline.
Big spatio-temporal datasets, available through both open and administrative data sources, offer significant potential for social science research. The magnitude of the data allows for increased resolution and analysis at individual level. While there are recent advances in forecasting techniques for highly granular te…
The paper explores how neural networks learn logical functions and their generalization error.
problem Learning logical functions with neural networks and understanding generalization error.
method Gradient descent on neural networks, analyzing noise-stability and Boolean influence.
result Gradient descent on certain neural architectures tends to favor low-degree representations, impacting generalization error.
Study shows annotation instrument design affects model performance in hate speech detection.
problem Impact of annotation instrument design on model performance in hate speech detection.
method Collected annotations from five experimental conditions of an annotation instrument, fine-tuned BERT models on each dataset, evaluated performance on holdout portion.
result Significant differences in model performance and annotations across conditions.
Identity-link IRT improves TVD-MI scores without curvature violations.
problem Preserving additivity in TVD-MI scores for efficient LLM evaluation.
method Derives clipped-linear model from Gini entropy maximization, using identity link.
result Identity-link yields lower curvature violations (median curl 0.080-0.150) compared to probit/logit.
New algorithms for model selection in off-policy evaluation of reinforcement learning.
problem Hyperparameter tuning for off-policy evaluation methods in reinforcement learning.
method Developed new model-free and model-based selectors with theoretical guarantees and a new experimental protocol.
result New model-free selector, LSTD-Tournament, demonstrates promising empirical performance.
Enhances model OOD detection by diversifying feature representation.
problem Limited diversity in model ensembles for OOD detection.
method Integrates diverse training tasks into model ensembles to generate distinct feature representations.
result Superior performance of Multi-Comprehension Ensemble in OOD detection.
Traditional statistical analysis requires that the analysis process and data are independent. By contrast, the new field of adaptive data analysis hopes to understand and provide algorithms and accuracy guarantees for research as it is commonly performed in practice, as an iterative process of interacting repeatedly wi…
The correct use of model evaluation, model selection, and algorithm selection techniques is vital in academic machine learning research as well as in many industrial settings. This article reviews different techniques that can be used for each of these three subtasks and discusses the main advantages and disadvantages …
Datasets are often reused to perform multiple statistical analyses in an adaptive way, in which each analysis may depend on the outcomes of previous analyses on the same dataset. Standard statistical guarantees do not account for these dependencies and little is known about how to provably avoid overfitting and false d…