A General Large Neighborhood Search Framework for Solving Integer Linear Programs
This addresses the challenge of improving optimization efficiency for combinatorial problems, offering a general approach that leverages existing solvers, though it is incremental as it builds on the LNS paradigm.
The paper tackles the problem of solving large-scale integer linear programs by proposing a data-driven large neighborhood search framework that learns to select neighborhoods using imitation and reinforcement learning, and demonstrates it significantly outperforms state-of-the-art commercial solvers like Gurobi in wall-clock time.
This paper studies a strategy for data-driven algorithm design for large-scale combinatorial optimization problems that can leverage existing state-of-the-art solvers in general purpose ways. The goal is to arrive at new approaches that can reliably outperform existing solvers in wall-clock time. We focus on solving integer programs, and ground our approach in the large neighborhood search (LNS) paradigm, which iteratively chooses a subset of variables to optimize while leaving the remainder fixed. The appeal of LNS is that it can easily use any existing solver as a subroutine, and thus can inherit the benefits of carefully engineered heuristic or complete approaches and their software implementations. We show that one can learn a good neighborhood selector using imitation and reinforcement learning techniques. Through an extensive empirical validation in bounded-time optimization, we demonstrate that our LNS framework can significantly outperform compared to state-of-the-art commercial solvers such as Gurobi.