Advances in Decision-Making under Uncertainty

The field of decision-making under uncertainty is rapidly advancing, with a focus on developing innovative methods to improve the accuracy and efficiency of decision-making processes. Researchers are exploring new approaches to address complex problems, such as multi-objective optimization, bandit optimization, and fair division. Notably, there is a growing interest in developing algorithms that can handle uncertain disruptions, adversarial attacks, and biased data. These advancements have significant implications for real-world applications, including online advertising, recommendation systems, and resource allocation. Some noteworthy papers in this area include: EGA, which proposes a unified end-to-end generative framework for industrial advertising systems, demonstrating its effectiveness in offline and online experiments. Best Arm Identification with Possibly Biased Offline Data, which introduces the LUCB-H algorithm to address the challenge of biased offline data in best arm identification problems, showing improved performance over standard LUCB algorithms.

Sources

EGA: A Unified End-to-End Generative Framework for Industrial Advertising Systems

Best Group Identification in Multi-Objective Bandits

Facility Location with Public Locations and Private Doubly-Peaked Costs

Enhanced Ideal Objective Vector Estimation for Evolutionary Multi-Objective Optimization

Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection

Online Fair Division for Personalized $2$-Value Instances

A Unified Online-Offline Framework for Co-Branding Campaign Recommendations

Continuum-armed Bandit Optimization with Batch Pairwise Comparison Oracles

Online Selection with Uncertain Disruption

Learning to Incentivize in Repeated Principal-Agent Problems with Adversarial Agent Arrivals

Best Arm Identification with Possibly Biased Offline Data

COBRA: Contextual Bandit Algorithm for Ensuring Truthful Strategic Agents

Built with on top of