Researchers proposed Collab-Solver, a novel multi-agent-based policy learning framework for mixed-integer linear programming (MILP). This framework enables collaborative policy optimization for multiple modules, leading to improved solving performance and generalization abilities. This development may be of interest to researchers and developers working on AI optimization techniques and their applications.

Researchers present LABELSENSE-Pilot, a reproducible prototype for optimal point-feature label placement on interactive maps, balancing geometric validity, display yield, and accessibility requirements. The system uses a multilayer perceptron to score layout candidates and ensures viewport containment, uniqueness, and clearance. Experiments demonstrate improved performance over a handcrafted integer program, with reduced flicker and display loss.

This research explores the application of reinforcement learning to adaptively couple full and reduced order models in hybrid simulations, particularly in transient problems where localized features propagate through the domain. The approach uses deep Q-networks to select between subdomain-local full order models and pre-trained operator inference reduced order models to balance accuracy, cost, and model-switching frequency.

Researchers introduced Monte Carlo Query Search (MCQS), a method for learning symbolic stochastic capability models of black-box AI agents. MCQS evaluates an agent's capabilities in novel settings by synthesizing queries that distinguish between optimistic and pessimistic models. Experiments show MCQS learns accurate models more efficiently than baseline strategies, enabling systematic characterization of agent capability boundaries.