On multiagent online problems with predictions

2025-07-15

AI Summary

This work studies online algorithm design with predictions for multi-agent decision-making, focusing on improving competitive ratios via two distinct predictors: a self-behavior predictor and an other-behavior predictor. We propose the first theoretical framework supporting dual predictors, rigorously distinguishing self- versus other-behavior modeling, and instantiate it on the multi-agent ski-rental problem. Under varying prediction quality assumptions, we characterize tight theoretical bounds on the optimal competitive ratio. Notably, when the other-behavior predictor is perfect, we prove that a self-prediction-only strategy achieves optimality. Furthermore, we design a new robust algorithm that significantly outperforms baseline approaches under imperfect predictions. Our work establishes a systematic analytical paradigm for prediction-augmented online algorithms in multi-agent settings, revealing the critical impact of predictive role division—between self and others—on both system performance and robustness.

Technology Category

• Multiagent Systems: Mechanism Design
• Machine Learning: Online Learning & Bandits
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

We study the power of (competitive) algorithms with predictions in a multiagent setting. We introduce a two predictor framework, that assumes that agents use one predictor for their future (self) behavior, and one for the behavior of the other players. The main problem we are concerned with is understanding what are the best competitive ratios that can be achieved by employing such predictors, under various assumptions on predictor quality. As an illustration of our framework, we introduce and analyze a multiagent version of the ski-rental problem. In this problem agents can collaborate by pooling resources to get a group license for some asset. If the license price is not met then agents have to rent the asset individually for the day at a unit price. Otherwise the license becomes available forever to everyone at no extra cost. In the particular case of perfect other predictions the algorithm that follows the self predictor is optimal but not robust to mispredictions of agent's future behavior; we give an algorithm with better robustness properties and benchmark it.

Problem

• Study competitive algorithms with predictions in multiagent settings
• Analyze best competitive ratios using dual-predictor framework
• Develop robust algorithms for multiagent ski-rental problem

Innovation

• Two predictor framework for multiagent predictions
• Optimal algorithm with self predictor
• Robust algorithm against mispredictions