Sample-Efficient Omniprediction for Proper Losses

• URL: http://arxiv.org/abs/2510.12769v1
• Date: Tue, 14 Oct 2025 17:49:05 GMT
• Title: Sample-Efficient Omniprediction for Proper Losses
• Authors: Isaac Gibbs, Ryan J. Tibshirani,
• Abstract summary: We consider the problem of constructing probabilistic predictions that lead to accurate decisions when employed by downstream users to inform actions.<n>For a single decision maker, designing an optimal predictor is equivalent to minimizing a proper loss function corresponding to the negative utility of that individual.<n>For multiple decision makers, our problem can be viewed as a variant of omniprediction in which the goal is to design a single predictor that simultaneously minimizes multiple losses.
• Score: 7.615840544913577
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We consider the problem of constructing probabilistic predictions that lead to accurate decisions when employed by downstream users to inform actions. For a single decision maker, designing an optimal predictor is equivalent to minimizing a proper loss function corresponding to the negative utility of that individual. For multiple decision makers, our problem can be viewed as a variant of omniprediction in which the goal is to design a single predictor that simultaneously minimizes multiple losses. Existing algorithms for achieving omniprediction broadly fall into two categories: 1) boosting methods that optimize other auxiliary targets such as multicalibration and obtain omniprediction as a corollary, and 2) adversarial two-player game based approaches that estimate and respond to the ``worst-case" loss in an online fashion. We give lower bounds demonstrating that multicalibration is a strictly more difficult problem than omniprediction and thus the former approach must incur suboptimal sample complexity. For the latter approach, we discuss how these ideas can be used to obtain a sample-efficient algorithm through an online-to-batch conversion. This conversion has the downside of returning a complex, randomized predictor. We improve on this method by designing a more direct, unrandomized algorithm that exploits structural elements of the set of proper losses.

Related papers

• Panprediction: Optimal Predictions for Any Downstream Task and Loss [22.59167297839645]
  Supervised learning is classically formulated as training a model to minimize a fixed loss function over a fixed distribution, or task.<n>We formalize a mathematical framework for this paradigm, which we call panprediction, and study its statistical complexity.<n>We show that simultaneously minimizing infinitely many losses on infinitely many tasks can be as statistically easy as minimizing one loss on one task.
  arXiv  Detail & Related papers  (2025-10-31T17:10:08Z)
• Online Decision-Focused Learning [74.3205104323777]
  Decision-focused learning (DFL) is an increasingly popular paradigm for training models whose predictive outputs are used in decision-making tasks.<n>In this paper, we regularize the objective function to make it different and investigate how to overcome nonoptimality function.<n>We also showcase the effectiveness of our algorithms on a knapsack experiment, where they outperform two standard benchmarks.
  arXiv  Detail & Related papers  (2025-05-19T10:40:30Z)
• Simple and Provable Scaling Laws for the Test-Time Compute of Large Language Models [70.07661254213181]
  We propose two algorithms that enjoy provable scaling laws for the test-time compute of large language models.<n>One is a two-stage knockout-style algorithm, where each candidate is evaluated by its average win rate against multiple opponents.<n>The other is a two-stage league-style algorithm, where each candidate is evaluated by its average win rate against multiple opponents.
  arXiv  Detail & Related papers  (2024-11-29T05:29:47Z)
• Refined Risk Bounds for Unbounded Losses via Transductive Priors [67.12679195076387]
  We revisit the sequential variants of linear regression with the squared loss, classification problems with hinge loss, and logistic regression.<n>Our key tools are based on the exponential weights algorithm with carefully chosen transductive priors.
  arXiv  Detail & Related papers  (2024-10-29T00:01:04Z)
• Decision-focused predictions via pessimistic bilevel optimization: complexity and algorithms [0.7499722271664147]
  We contribute to recent efforts in decision-focused predictions, i.e., to build predictive models.<n>We show computational results of a problem, including its membership in NP.<n>We then formulate various computational techniques to achieve empirical tractability.
  arXiv  Detail & Related papers  (2023-12-29T15:05:00Z)
• High-Dimensional Prediction for Sequential Decision Making [9.684829814477526]
  We study the problem of making predictions of an adversarially chosen high-dimensional state that are unbiased subject to an arbitrary collection of conditioning events. We give efficient algorithms for solving this problem, as well as a number of applications that stem from choosing an appropriate set of conditioning events.
  arXiv  Detail & Related papers  (2023-10-26T17:59:32Z)
• Semantic Strengthening of Neuro-Symbolic Learning [85.6195120593625]
  Neuro-symbolic approaches typically resort to fuzzy approximations of a probabilistic objective. We show how to compute this efficiently for tractable circuits. We test our approach on three tasks: predicting a minimum-cost path in Warcraft, predicting a minimum-cost perfect matching, and solving Sudoku puzzles.
  arXiv  Detail & Related papers  (2023-02-28T00:04:22Z)
• A Note on Task-Aware Loss via Reweighing Prediction Loss by Decision-Regret [11.57423546614283]
  We propose a decision-aware version of predict-then-optimize. We reweigh the prediction error by the decision regret incurred by an (unweighted) pilot estimator of costs. We show that this approach can lead to improvements over a "predict-then-optimize" framework.
  arXiv  Detail & Related papers  (2022-11-09T18:59:35Z)
• Efficient and Differentiable Conformal Prediction with General Function Classes [96.74055810115456]
  We propose a generalization of conformal prediction to multiple learnable parameters. We show that it achieves approximate valid population coverage and near-optimal efficiency within class. Experiments show that our algorithm is able to learn valid prediction sets and improve the efficiency significantly.
  arXiv  Detail & Related papers  (2022-02-22T18:37:23Z)
• The Perils of Learning Before Optimizing [16.97597806975415]
  We show how prediction models can be learned end-to-end by differentiating through the optimization task. We show that the performance gap between a two-stage and end-to-end approach is closely related to the emphprice of correlation concept in optimization.
  arXiv  Detail & Related papers  (2021-06-18T20:43:47Z)
• Multivariate Probabilistic Regression with Natural Gradient Boosting [63.58097881421937]
  We propose a Natural Gradient Boosting (NGBoost) approach based on nonparametrically modeling the conditional parameters of the multivariate predictive distribution. Our method is robust, works out-of-the-box without extensive tuning, is modular with respect to the assumed target distribution, and performs competitively in comparison to existing approaches.
  arXiv  Detail & Related papers  (2021-06-07T17:44:49Z)
• Universal Online Convex Optimization Meets Second-order Bounds [74.0120666722487]
  We propose a simple strategy for universal online convex optimization. The key idea is to construct a set of experts to process the original online functions, and deploy a meta-algorithm over the linearized losses. In this way, we can plug in off-the-shelf online solvers as black-box experts to deliver problem-dependent regret bounds.
  arXiv  Detail & Related papers  (2021-05-08T11:43:49Z)
• Divide and Learn: A Divide and Conquer Approach for Predict+Optimize [50.03608569227359]
  The predict+optimize problem combines machine learning ofproblem coefficients with a optimization prob-lem that uses the predicted coefficients. We show how to directlyexpress the loss of the optimization problem in terms of thepredicted coefficients as a piece-wise linear function. We propose a novel divide and algorithm to tackle optimization problems without this restriction and predict itscoefficients using the optimization loss.
  arXiv  Detail & Related papers  (2020-12-04T00:26:56Z)

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.