![[2410.08958] The MAPS Algorithm: Fast model-agnostic and distribution-free prediction intervals for supervised learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2410.08958] The MAPS Algorithm: Fast model-agnostic and distribution-free prediction intervals for supervised learning
Summary
The MAPS algorithm offers a novel approach to generating model-agnostic, distribution-free prediction intervals in supervised learning, addressing high-dimensional data challenges.
Why It Matters
This research is significant as it tackles the limitations of existing methods in producing reliable prediction intervals, which are crucial for understanding model uncertainty. By introducing a bootstrap-based algorithm that adapts to various predictive models, it enhances the robustness and applicability of predictive analytics in diverse fields, including image classification and simulation-based inference.
Key Takeaways
- The MAPS algorithm provides distribution-free conditional prediction intervals.
- It is model-agnostic and adapts to any trained predictive model.
- The algorithm addresses high-dimensional input challenges and heteroscedastic errors.
- It connects prediction accuracy with interval length and ensures asymptotic conditional coverage.
- MAPS is applicable in both simulation-based inference and image classification.
Statistics > Machine Learning arXiv:2410.08958 (stat) [Submitted on 11 Oct 2024 (v1), last revised 20 Feb 2026 (this version, v2)] Title:The MAPS Algorithm: Fast model-agnostic and distribution-free prediction intervals for supervised learning Authors:Daniel Salnikov, Dan Leonte, Kevin Michalewicz View a PDF of the paper titled The MAPS Algorithm: Fast model-agnostic and distribution-free prediction intervals for supervised learning, by Daniel Salnikov and Dan Leonte and Kevin Michalewicz View PDF HTML (experimental) Abstract:A fundamental problem in modern supervised learning is computing reliable conditional prediction intervals in high-dimensional settings: existing methods often rely on restrictive modelling assumptions, do not scale as predictor dimension increases, or only guarantee marginal (population-level) rather than conditional (individual-level) coverage. We introduce the $\textit{lifted predictive model}$ (LPM), a new conditional representation, and propose the MAPS (Model-Agnostic Prediction Sets) algorithm that produces distribution-free conditional prediction intervals and adapts to any trained predictive model. Our procedure is bootstrap-based, scales to high-dimensional inputs and accounts for heteroscedastic errors. We establish the theoretical properties of the LPM, connect prediction accuracy to interval length, and provide sufficient conditions for asymptotic conditional coverage. We evaluate the finite-sample performance of MAPS in a simulation stud...
