analiza aiqbench
InvEvolve introduces a framework that leverages reinforcement learning and confidence-interval-based certification to generate inventory policies with provable statistical safety guarantees for non-stationary demand. By connecting training, inference, and deployment through theoretical bounds, the system outperforms both classical methods and standard deep-learning approaches on real-world retail data. This work is significant for practitioners seeking to deploy LLMs in dynamic operational settings where reliability and performance verification are critical.
InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees
arXiv:2605.00369v4 Announce Type: replace Abstract: We study how large language models can be used to generate inventory policies in online settings with non-stationary demand. Our work is motivated by recent advances in LLM-based evolutionary search, such as AlphaEvolve, which demonstrates strong performance on static and highly structured problems such as mathematical discovery, but is not directly suited to dynamic inventory settings with online updates. We propose InvEvolve, an end-to-end inventory policy evolution and inference framework grounded in confidence-interval-based certification. Built on a large language model trained via reinforcement learning, InvEvolve can process demand data together with additional numerical and textual features, and generates white-box inventory policies with statistical safety guarantees for future deployment. We further introduce a unified framework with theoretical guarantees that connects training, inference, and deployment. This allows us to derive a lower bound on the probability that InvEvolve evolves a statistically safe and improved policy, and to characterize the multi-period performance gap relative to the oracle-safe benchmark. Tested on both synthetic data and real-world retail data, InvEvolve outperforms classical inventory policies and deep-learning-based methods. In canonical inventory settings, it generates new policies that outperform existing benchmarks.
Abstract:We study how large language models can be used to generate inventory policies in online settings with non-stationary demand. Our work is motivated by recent advances in LLM-based evolutionary search, such as AlphaEvolve, which demonstrates strong performance on static and highly structured problems such as mathematical discovery, but is not directly suited to dynamic inventory settings with online updates. We propose InvEvolve, an end-to-end inventory policy evolution and inference framework grounded in confidence-interval-based certification. Built on a large language model trained via reinforcement learning, InvEvolve can process demand data together with additional numerical and textual features, and generates white-box inventory policies with statistical safety guarantees for future deployment. We further introduce a unified framework with theoretical guarantees that connects training, inference, and deployment. This allows us to derive a lower bound on the probability that InvEvolve evolves a statistically safe and improved policy, and to characterize the multi-period performance gap relative to the oracle-safe benchmark. Tested on both synthetic data and real-world retail data, InvEvolve outperforms classical inventory policies and deep-learning-based methods. In canonical inventory settings, it generates new policies that outperform existing benchmarks.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2605.00369 [cs.LG]
(or arXiv:2605.00369v4 [cs.LG] for this version)
Text integral sub licența deschisă a publicației · 198 cuvinte
Subiecte conexe
Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification
arXiv · 1 oct. 2026
The Value of Human Expertise
arXiv · 27 aug. 2026
Adaptive-Horizon Conflict-Based Search for Closed-Loop Multi-Agent Path Finding
arXiv · 12 iun. 2026
General Performance Guarantee for Human Torque Estimation-Based Task-Agnostic Assistive Exoskeleton Control
arXiv · 1 oct. 2026