A Tale of Two Variances: When Single-Seed Benchmarks Fail in Bayesian Deep Learning

• URL: http://arxiv.org/abs/2604.23114v1
• Date: Sat, 25 Apr 2026 02:52:33 GMT
• Title: A Tale of Two Variances: When Single-Seed Benchmarks Fail in Bayesian Deep Learning
• Abstract summary: Local CRPS variance provides a direct signal of single-seed estimation error.<n>Power-law fit quality and monotonicity together provide compact method-level summaries of trajectory regularity.
• Score: 10.891872788947298
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In limited-data settings, a single endpoint mean of an evaluation metric such as the Continuous Ranked Probability Score (CRPS) is itself a random variable, yet it is routinely reported as if it were a stable property of the method. We study when this practice fails. Using 50 independent repetitions across six regression datasets, we show that CRPS variance trajectories differ substantially across methods and are not always well described by a smooth power-law decay. Methods with a learned heteroscedastic variance head, namely MAP and Deep Ensembles, can develop pronounced, reproducible variance peaks at intermediate training sizes on real datasets, whereas MC Dropout and Bayes by Backprop typically show smooth variance contraction. These peaks have direct practical consequences: at the variance peak on Seoul Bike, the relative RMSE of a single-seed MAP estimate reaches 93.6\%, and the probability of falling within \(\pm 10\%\) of the repeated-run mean drops to 5.9\%. We show that local CRPS variance provides a direct signal of single-seed estimation error, with Spearman correlations above 0.96 on every real dataset. Power-law fit quality and monotonicity together provide compact method-level summaries of trajectory regularity. Finally, replacing the standard heteroscedastic objective with \(β\)-NLL substantially reduces the irregular behavior, consistent with the view that the heteroscedastic training objective contributes to the instability. Practitioners should report trajectory summaries alongside endpoint means and concentrate repeated evaluation in high-variance regions.

Related papers

• Semiparametric Inference for Conditional Shapley Feature Importance [0.0]
  This paper studies the conditional formulation, in which out-of-coalition features are integrated out under their true conditional distribution.<n>We propose a one-step estimator with K-fold cross-fitting and a U-statistic correction of the squared loss interval.<n>A Pinsker-type bound quantifies the bias from misspecifying the working copula class, while vine copulas keep conditional sampling tractable.
  arXiv  Detail & Related papers  (2026-09-09T15:19:24Z)
• Beyond Homoscedasticity: Decoupled Uncertainty Optimization for Deep Imbalanced Regression [7.011984503405774]
  Deep Imbalanced Regression is pervasive in continuous prediction tasks across diverse modalities.<n>We propose DUO, an uncertainty-aware long-tailed regression framework.<n>We show that DUO achieves the best few-shot bMAE and GM on IMDB-WIKI-DIR, AgeDB-DIR, and AAV2-DIR while remaining competitive on few-shot MAE.
  arXiv  Detail & Related papers  (2026-09-04T11:06:57Z)
• Convex-Hull-Neighborhood Smooth Dual Generalization: Controlling Local Correction Propagation in Offline RL [7.803284875836737]
  bootstrapping can amplify estimation errors in offline reinforcement learning.<n>We propose Convex Hull Neighborhood Smooth Dual Generalization, which expresses the Bellman backup as an in-sample value target plus a CHN-local correction.<n> Experiments on Gym-MuJoCo and AntMaze show strong aggregate performance and stable value estimation.
  arXiv  Detail & Related papers  (2026-08-04T04:30:02Z)
• Distributionally Faithful Imputation via Positive Semi-Definite Kernel Density Estimation [55.47411235954028]
  We recast imputation under missing completely at random as density estimation from masked observations.<n>We obtain a convex empirical risk problem with closed form marginals, solvable by a Newton interior point method.<n>The resulting PSD Impute model yields both single and multiple imputations from the same fitted density.
  arXiv  Detail & Related papers  (2026-07-08T15:51:05Z)
• Spatially Robust Inference with Predicted and Missing at Random Labels [2.5574009994099196]
  We show that cross-fitting induces fold-level correlation that distorts spatial variance estimators, producing unstable or overly conservative confidence intervals.<n>We also propose a jackknife spatial heterosasticity and auto consistent (HAC) variance correction that separates dependence from fold-induced noise.
  arXiv  Detail & Related papers  (2026-03-11T23:14:21Z)
• Computable Bernstein Certificates for Cross-Fitted Clipped Covariance Estimation [0.0]
  We propose a cross-fitted clipped covariance estimator equipped with emphfully computable Bernstein-type deviation certificates.<n>The resulting procedure adapts to intrinsic complexity measures such as effective rank under mild tail regularity.
  arXiv  Detail & Related papers  (2026-02-15T06:53:40Z)
• Ensemble Threshold Calibration for Stable Sensitivity Control [0.0]
  We present an end-to-end framework that achieves exact recall with sub-percent variance over tens of millions of geometry pairs.<n>Our approach consistently hits a recall target within a small error, decreases redundant verifications relative to other calibrations, and runs end-to-end on a single TPU v3 core.
  arXiv  Detail & Related papers  (2025-10-02T15:22:28Z)
• Towards Self-Supervised Covariance Estimation in Deep Heteroscedastic Regression [102.24287051757469]
  We study self-supervised covariance estimation in deep heteroscedastic regression.<n>We derive an upper bound on the 2-Wasserstein distance between normal distributions.<n>Experiments over a wide range of synthetic and real datasets demonstrate that the proposed 2-Wasserstein bound coupled with pseudo label annotations results in a computationally cheaper yet accurate deep heteroscedastic regression.
  arXiv  Detail & Related papers  (2025-02-14T22:37:11Z)
• Aioli: A Unified Optimization Framework for Language Model Data Mixing [74.50480703834508]
  We show that no existing method consistently outperforms a simple stratified sampling baseline in terms of average test perplexity.<n>We derive a new online method named Aioli, which directly estimates the mixing law parameters throughout training and uses them to dynamically adjust proportions.
  arXiv  Detail & Related papers  (2024-11-08T17:50:24Z)
• Geometry-Aware Instrumental Variable Regression [56.16884466478886]
  We propose a transport-based IV estimator that takes into account the geometry of the data manifold through data-derivative information. We provide a simple plug-and-play implementation of our method that performs on par with related estimators in standard settings.
  arXiv  Detail & Related papers  (2024-05-19T17:49:33Z)
• An analysis of the noise schedule for score-based generative models [7.180235086275926]
  Score-based generative models (SGMs) aim at estimating a target data distribution by learning score functions using only noise-perturbed samples from the target.<n>Recent literature has focused extensively on assessing the error between the target and estimated distributions, gauging the generative quality through the Kullback-Leibler (KL) divergence and Wasserstein distances.<n>We establish an upper bound for the KL divergence between the target and the estimated distributions, explicitly depending on any time-dependent noise schedule.
  arXiv  Detail & Related papers  (2024-02-07T08:24:35Z)
• TIC-TAC: A Framework for Improved Covariance Estimation in Deep Heteroscedastic Regression [109.69084997173196]
  Deepscedastic regression involves jointly optimizing the mean and covariance of the predicted distribution using the negative log-likelihood. Recent works show that this may result in sub-optimal convergence due to the challenges associated with covariance estimation. We study two questions: (1) Does the predicted covariance truly capture the randomness of the predicted mean? Our results show that not only does TIC accurately learn the covariance, it additionally facilitates an improved convergence of the negative log-likelihood.
  arXiv  Detail & Related papers  (2023-10-29T09:54:03Z)
• Policy evaluation from a single path: Multi-step methods, mixing and mis-specification [45.88067550131531]
  We study non-parametric estimation of the value function of an infinite-horizon $gamma$-discounted Markov reward process. We provide non-asymptotic guarantees for a general family of kernel-based multi-step temporal difference estimates.
  arXiv  Detail & Related papers  (2022-11-07T23:15:25Z)
• Benign Overfitting of Constant-Stepsize SGD for Linear Regression [122.70478935214128]
  inductive biases are central in preventing overfitting empirically. This work considers this issue in arguably the most basic setting: constant-stepsize SGD for linear regression. We reflect on a number of notable differences between the algorithmic regularization afforded by (unregularized) SGD in comparison to ordinary least squares.
  arXiv  Detail & Related papers  (2021-03-23T17:15:53Z)
• Attentional-Biased Stochastic Gradient Descent [74.49926199036481]
  We present a provable method (named ABSGD) for addressing the data imbalance or label noise problem in deep learning. Our method is a simple modification to momentum SGD where we assign an individual importance weight to each sample in the mini-batch. ABSGD is flexible enough to combine with other robust losses without any additional cost.
  arXiv  Detail & Related papers  (2020-12-13T03:41:52Z)
• GenDICE: Generalized Offline Estimation of Stationary Values [108.17309783125398]
  We show that effective estimation can still be achieved in important applications. Our approach is based on estimating a ratio that corrects for the discrepancy between the stationary and empirical distributions. The resulting algorithm, GenDICE, is straightforward and effective.
  arXiv  Detail & Related papers  (2020-02-21T00:27:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.