Derives ideal train/test split for ridge regression in large data limit.
problem Finding optimal train/test split for ridge regression in large data scenarios.
method Mathematical derivation of optimal train/test split, considering ridge tuning parameter and asymptotic behavior.
result The optimal train/test split for ridge regression in the large data limit depends weakly on the ridge tuning parameter alpha.
Optimal data split ratio is sqrt(p):1 for linear regression.
problem Lack of clear guidance on optimal training/testing data split ratio.
method Showed that optimal ratio is sqrt(p):1 for linear regression.
result Optimal ratio for training/testing split is sqrt(p):1.
Study robustness of split conformal prediction under adversarial attacks.
problem Ensuring distribution-free coverage guarantees in CP under adversarial conditions.
method Theoretical analysis and extensive experiments on split conformal prediction robustness.
result Prediction coverage varies with calibration-time attack strength, enabling control over coverage under adversarial tests.
SPlit optimizes dataset splitting for better model performance.
problem Improving model performance through optimal dataset splitting.
method Adapting Support Points (SP) algorithm for subsampling and categorical variables in a sequential nearest neighbor approach.
result SPlit significantly improves worst-case testing performance compared to random splitting.
Develops a new inference method for split-sample estimators using multiple splits.
problem Statistical dependence and variability in split-sample estimators.
method Averaging across multiple splits, proving a central limit theorem, and developing new inference approaches.
result Valid confidence intervals and improved power in comparing model performance.
SBSS uses similarity to split data for better classifier training.
problem Training better classifiers with realistic performance estimation.
method SBSS uses both input and output space information to split data using similarity functions.
result SBSS outperformed ordinary stratified 10-fold cross-validation in 75% of scenarios.
In this study, we applied a stochastic spread pairs trading strategy on the Indian commodity market. The complete set of commodities were taken whose spot price was available for the period of January 1st 2010 to December 31st 2018 including energy, metals and the agricultural commodity sector. Spot data was taken from…
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.
A new method for kernel tests without data splitting increases power.
problem Lack of power in kernel-based tests due to data splitting.
method Selective inference framework to learn hyperparameters and test on full sample.
result Empirically larger test power without data splitting, regardless of split proportion.
Meta-learning performance depends on train-validation split type.
problem Understanding the importance of train-validation split in meta-learning.
method Theoretical and experimental study comparing train-val and train-train methods.
result Train-train method can achieve strictly better excess loss in realizable cases.
Twinning splits data into fast, statistically similar sets.
problem Creating statistically similar data splits for Big Data.
method Twinning is a method based on SPlit for fast, model-independent dataset splitting.
result Twinning is orders of magnitude faster than SPlit.
Proposes a new cross-validation method to estimate model performance.
problem The standard cross-validation method does not accurately estimate the performance of the recommended model.
method Develops a new random-effects model framework to improve naive cross-validation estimators.
result Proposed estimators outperform conventional and naive methods in estimating model performance.
E-C2ST uses E-values for high-dimensional data two-sample tests.
problem Statistical testing for high-dimensional data.
method Combines split likelihood ratio tests and predictive independence tests, using E-values for anytime-valid sequential tests.
result E-C2ST achieves enhanced statistical power by partitioning datasets into multiple batches.
Paper uses Random Matrix Theory for optimal training-testing data split.
problem Finding ideal training-testing data split for linear regression.
method Random Matrix Theory applied to Gaussian multivariate data.
result Ideal training and test sizes derived for any model.
SVHN dataset's split affects generative models but not digit classification.
problem Distribution mismatch between SVHN training and test sets impacts generative models.
method Empirically showed distribution mismatch affects generative models; proposed mixing and re-splitting.
result Distribution mismatch in SVHN dataset significantly impacts probabilistic generative models.
A machine learning configuration refers to a combination of preprocessor, learner, and hyperparameters. Given a set of configurations and a large dataset randomly split into training and testing set, we study how to efficiently select the best configuration with approximately the highest testing accuracy when trained f…
Develops significance tests for neural networks without strong assumptions or excessive computation.
problem Addressing the black-box nature of deep neural networks for feature relevance testing.
method Derives one-split and two-split tests relaxing assumptions and computational complexity.
result Establishes asymptotic null distributions and consistency in Type II error.
SOAK assesses data subset similarity for better model training.
problem Estimating similarity between data subsets for accurate predictions.
method Same/Other/All K-fold cross-validation method.
result SOAK estimates similarity of learnable/predictable patterns in data subsets.
Memory split advantage: thinner networks outperform a single wide network.
problem Optimizing deep learning models with limited memory.
method Investigated training a single wide network vs. an ensemble of thinner networks with the same total number of parameters.
result An ensemble of several thinner networks outperforms a single wide network for large memory budgets.
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.
USNRT uses tree-structured learning to improve uncertainty quantification of variance networks.
problem Improving uncertainty quantification of variance networks.
method Tree-structured local neural network model that partitions feature space into regions for training region-specific neural networks to predict mean and variance.
result USNRT shows superior performance in estimating uncertainty with variances on UCI datasets compared to recent methods.
We present Fashion-MNIST, a new dataset comprising of 28x28 grayscale images of 70,000 fashion products from 10 categories, with 7,000 images per category. The training set has 60,000 images and the test set has 10,000 images. Fashion-MNIST is intended to serve as a direct drop-in replacement for the original MNIST dat…
A new method for feature importance inference without data splitting.
problem Feature importance inference for machine learning models.
method Minipatch ensembles for model-agnostic, distribution-free inference.
result Asymptotic validity of confidence intervals without data splitting.
New method combines randomization tests and flexible models for valid inference without splitting data.
problem Valid inference in randomized panel experiments with complex effect heterogeneity.
method Model-assisted randomization tests that estimate unsigned CATE from residualized outcomes.
result CATE-assisted tests control Type I error and achieve higher power than alternatives.
This paper explores how train-validation splits help in NAS to prevent overfitting.
problem NAS overfits with train-validation splits and needs better generalization guarantees.
method Established refined properties of validation loss and risk for NAS.
result NAS with train-validation splits can select the most generalizable model.
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.
Unified data representation learning improves non-parametric two-sample testing.
problem Improving non-parametric two-sample testing accuracy.
method Proposes RL-TST framework combining IRs and DRs for better test power.
result RL-TST outperforms existing methods by leveraging both IRs and DRs.
Study robustness of split conformal prediction in data contamination setting.
problem Robustness of split conformal prediction under data contamination.
method Analyze split conformal prediction's performance in a contaminated data setting and propose a new method.
result Demonstrated the impact of corrupted data on prediction intervals' coverage and efficiency.
Paper analyzes biases in video QA datasets, showing models can answer 37-48% questions correctly without multimodal context.
problem Question answering biases in video QA datasets can lead to model overfitting and poor generalization.
method Analyzed popular video question answering datasets, conducted ablation studies on biases from annotators and question types.
result Pretrained language models can answer 37-48% questions correctly without multimodal context, far exceeding random guess baseline.
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.
In recent years, many non-traditional classification methods, such as Random Forest, Boosting, and neural network, have been widely used in applications. Their performance is typically measured in terms of classification accuracy. While the classification error rate and the like are important, they do not address a fun…
This paper introduces a Decision Tree Learner as an early warning system for classification of the non-life insurance companies according to their financial solid as strong, moderate, weak, or insolvency. In this study, we ran several experiments to show that the proposed model can achieve a good result using standard …
A variety of machine learning tasks---e.g., matrix factorization, topic modelling, and feature allocation---can be viewed as learning the parameters of a probability distribution over bipartite graphs. Recently, a new class of models for networks, the sparse exchangeable graphs, have been introduced to resolve some imp…
Tests factor models by decomposing market into body and tail legs, revealing inconsistent results.
problem Inconsistency between factor models and market behavior.
method Decomposes market into body and tail legs, testing factor models at daily and monthly frequencies.
result q5 model shows inconsistent results, with negative body and positive tail alphas at all split ratios.
LoBoost improves local conformal prediction for gradient-boosted trees without extra data splits.
problem Quantifying uncertainty in gradient-boosted tree predictions.
method Model-native local conformal prediction using leaf structure.
result Competitive interval quality and improved test MSE with large calibration speedups.
A new MMD-based test combines kernels for two-sample testing without splitting data.
problem Efficiently testing if two datasets come from the same distribution without splitting data.
method Proposes a novel statistic based on Maximum Mean Discrepancy (MMD) that combines kernels, proving concentration bounds and showing data-dependent kernel selection.
result Exponential concentration bounds and improved test power compared to existing methods.
This study examines how randomness affects machine learning model performance.
problem The impact of randomness on machine learning model performance.
method Empirical study comparing randomness in model training and dataset partitioning.
result Randomness in model training causes more variation in FFNNs than tree-based methods.
Semi-supervised node classification in graphs is a fundamental problem in graph mining, and the recently proposed graph neural networks (GNNs) have achieved unparalleled results on this task. Due to their massive success, GNNs have attracted a lot of attention, and many novel architectures have been put forward. In thi…
The recent success of generative adversarial networks and variational learning suggests training a classifier network may work well in addressing the classical two-sample problem. Network-based tests have the computational advantage that the algorithm scales to large samples. This paper proposes a two-sample statistic …
Machine learning experiments show IID assumption is flawed for bathymetry editing.
problem Flawed IID assumption in machine learning for bathymetry editing.
method Real-world computer-assisted labeling task, IID assumption analysis.
result Common random split leads to poor performance in machine learning.
We propose a general method for constructing hypothesis tests and confidence sets that have finite sample guarantees without regularity conditions. We refer to such procedures as "universal." The method is very simple and is based on a modified version of the usual likelihood ratio statistic, that we call "the split li…
HATT improves online decision tree ensembles by using a more eager splitting strategy.
problem Improving the efficiency of online decision tree ensembles.
method Replacing Hoeffding Tree's split strategy with HATT, which uses the Hoeffding Test for candidate splits.
result HATT outperforms Hoeffding Tree in online bagging and boosting ensembles, as shown by significant performance improvements in various testbenches.
Noisy labels are ubiquitous in real-world datasets, which poses a challenge for robustly training deep neural networks (DNNs) as DNNs usually have the high capacity to memorize the noisy labels. In this paper, we find that the test accuracy can be quantitatively characterized in terms of the noise ratio in datasets. In…
This paper examines the stability of learned explanations for black-box predictions via model distillation with decision trees. One approach to intelligibility in machine learning is to use an understandable `student' model to mimic the output of an accurate `teacher'. Here, we consider the use of regression trees as a…
KSDAgg combines multiple KSD tests to improve goodness-of-fit testing without splitting data.
problem Improving goodness-of-fit testing without data splitting.
method KSDAgg aggregates multiple KSD tests with different kernels to maximize power.
result KSDAgg achieves the smallest uniform separation rate of the collection, up to a logarithmic term.
Conventional decision trees have a number of favorable properties, including interpretability, a small computational footprint and the ability to learn from little training data. However, they lack a key quality that has helped fuel the deep learning revolution: that of being end-to-end trainable, and to learn from scr…
Ensembles, where multiple neural networks are trained individually and their predictions are averaged, have been shown to be widely successful for improving both the accuracy and predictive uncertainty of single neural networks. However, an ensemble's cost for both training and testing increases linearly with the numbe…
This paper presents the asymptotic behavior of a linear instrumental variables (IV) estimator that uses a ridge regression penalty. The regularization tuning parameter is selected empirically by splitting the observed data into training and test samples. Conditional on the tuning parameter, the training sample creates …