VLR: A Memory-based Optimization Heuristic

Abstract

We suggest a novel memory-based metaheuristic optimization algorithm, VLR, which uses a list of already-visited areas to more effectively search for an optimal solution. We chose the Max-cut problem to test its optimization performance, comparing it with state-of-the-art methods. VLR dominates the previous best-performing heuristics. We also undertake preliminary analysis of the algorithm’s parameter space, noting that a larger memory improves performance. VLR was designed as a general-purpose optimization algorithm, so its performance on other prob lems will be investigated in future.

Myoung Hoon Ha

Postdoctoral Researcher

My research interests include deep learning, reinforcement learning, representation learning, and adversarial robustness.