High-Dimensional Prediction for Sequential Decision Making

Georgy Noarov, Ramya Ramalingam, Aaron Roth, Stephan Xie

International Conference on Machine Learning 2025 · Oral

This talk introduces a novel framework for **high-dimensional prediction in sequential decision-making**, particularly in online adversarial environments. Presented by Ramya Ramalingam, a PhD student at the University of Pennsylvania, alongside co-authors Georgy Noarov, Aaron Roth, and Stephan Xie, the work addresses a fundamental challenge in modern machine learning systems: how to generate predictions that are not only accurate but also *trustworthy* and *useful* for downstream agents making critical decisions based on these forecasts. The core innovation lies in moving beyond traditional notions of **calibration**, which become computationally and statistically intractable in high-dimensional settings, towards a more generalized and efficient concept termed **event unbiasedness**.

AI review

A rigorous theoretical contribution from the Roth group at Penn that introduces 'event unbiasedness' as a tractable surrogate for high-dimensional calibration in online adversarial settings. The core result — a bias bound scaling as O(log(D·|F|)·√incidence) with polynomial runtime in the event family size — is clean, non-trivial, and demonstrably useful. The framework unifies several previously disconnected guarantees (conditional regret, multi-calibration, swap regret for multiple agents) under a single abstraction, which is exactly the kind of structural insight that earns citations. The…