The field of decision making under uncertainty is rapidly advancing, with a growing focus on learning-augmented approaches. Recent research has explored the integration of machine learning predictions into decision-making frameworks, enabling more accurate and robust outcomes. A key direction in this area is the development of mechanisms that can effectively leverage imperfect predictions to improve decision quality, while also ensuring robustness to prediction errors. Notable papers in this regard include 'Learning to Ask: Decision Transformers for Adaptive Quantitative Group Testing', which demonstrates the potential of adaptive algorithms to reduce the number of queries required in group testing problems, and 'AdaSwitch: An Adaptive Switching Meta-Algorithm for Learning-Augmented Bounded-Influence Problems', which introduces a meta-algorithm for online decision-making problems with sequence-based predictions. Overall, the field is moving towards the development of more sophisticated and adaptive decision-making frameworks that can effectively harness the power of machine learning predictions.
Advances in Learning-Augmented Decision Making
Sources
Strategyproof Mechanisms for Facility Location with Prediction Under the Maximum Cost Objective
Learning to Ask: Decision Transformers for Adaptive Quantitative Group Testing
Threshold-Based Optimal Arm Selection in Monotonic Bandits: Regret Lower Bounds and Algorithms
AdaSwitch: An Adaptive Switching Meta-Algorithm for Learning-Augmented Bounded-Influence Problems
Simulating classification models to evaluate Predict-Then-Optimize methods
Online Learning of Optimal Sequential Testing Policies
Drift Plus Optimistic Penalty -- A Learning Framework for Stochastic Network Optimization
Mistake-bounded online learning with operation caps
Smoothed Online Optimization for Target Tracking: Robust and Learning-Augmented Algorithms
Knapsack Contracts and the Importance of Return-on-Investment
The Keychain Problem: On Minimizing the Opportunity Cost of Uncertainty
Edge Server Monitoring for Job Assignment
Admission Control for Inelastic Traffic on a Link Shared by Deadline-Driven Elastic Traffic
Prediction Loss Guided Decision-Focused Learning