aiqbench analysis
This paper introduces CVaR-RMD, a risk-sensitive framework that leverages large language models to evolve robust heuristics by optimizing for worst-case performance rather than just average metrics. By using conditional value-at-risk as a selection pressure, the method significantly reduces failure rates across diverse combinatorial optimization domains while maintaining competitive mean performance. This approach addresses a critical limitation in automated algorithm design where standard mean-based objectives often obscure rare but severe degradation scenarios.
CVaR-RMD: CVaR Selection Pressure for Robust Metaheuristic Discovery
arXiv:2610.25697v1 Announce Type: new Abstract: Large language models have recently been used to discover heuristics through iterative program search, offering a promising approach to automated algorithm design. However, candidates selected by average performance may degrade across instance families, random seeds, and problem scales because mean-based objectives can obscure rare but severe failures. We propose CVaR-RMD, a risk-sensitive evolutionary framework that uses an LLM to mutate executable heuristic programs and ranks candidates by empirical conditional value-at-risk over heterogeneous evaluation conditions. Across three combinatorial optimization domains, CVaR-based selection improves upper-tail performance and reduces failure rates while maintaining competitive mean performance. These gains persist on larger unseen instances and independently curated within-domain benchmarks
Abstract:Large language models have recently been used to discover heuristics through iterative program search, offering a promising approach to automated algorithm design. However, candidates selected by average performance may degrade across instance families, random seeds, and problem scales because mean-based objectives can obscure rare but severe failures. We propose CVaR-RMD, a risk-sensitive evolutionary framework that uses an LLM to mutate executable heuristic programs and ranks candidates by empirical conditional value-at-risk over heterogeneous evaluation conditions. Across three combinatorial optimization domains, CVaR-based selection improves upper-tail performance and reduces failure rates while maintaining competitive mean performance. These gains persist on larger unseen instances and independently curated within-domain benchmarks
Accepted to Findings of EMNLP 2026. Code: this https URL
Artificial Intelligence (cs.AI)
arXiv:2610.25697 [cs.AI]
(or arXiv:2610.25697v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.25697 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 207 words
Related coverage
Inner optimizers in active-learning Kriging reliability analysis: a pre-specified comparison and the unconservative estimates left by unlearned failure modes
arXiv · Nov 3, 2026
EMG-Hybrid: A Dual-Engine Elite Memory--Guided Adaptive Optimization Framework for High-Dimensional Feature Selection
arXiv · Nov 4, 2026
Drive, Pack, Fly: The Travelling Thief Problem with Drone
arXiv · Aug 18, 2026
Reinforcement Learning-Guided Evolutionary Policy Optimization for Preference-Adjustable Heterogeneous Agile Earth Observation Satellite Scheduling
arXiv · Aug 27, 2026