Self-Labeling the Job Shop Scheduling Problem
This work addresses the need for expensive optimal solutions in combinatorial optimization, offering a method that eliminates reliance on exact solvers, though it is incremental as it builds on existing generative models like Pointer Networks.
The authors tackled the problem of costly target solutions in supervised learning for combinatorial problems by proposing a self-supervised training strategy called SLIM, which uses generative models to sample solutions and pseudo-label the best one, resulting in models that outperform constructive heuristics and state-of-the-art learning proposals for the Job Shop Scheduling Problem.
This work proposes a self-supervised training strategy designed for combinatorial problems. An obstacle in applying supervised paradigms to such problems is the need for costly target solutions often produced with exact solvers. Inspired by semi- and self-supervised learning, we show that generative models can be trained by sampling multiple solutions and using the best one according to the problem objective as a pseudo-label. In this way, we iteratively improve the model generation capability by relying only on its self-supervision, eliminating the need for optimality information. We validate this Self-Labeling Improvement Method (SLIM) on the Job Shop Scheduling (JSP), a complex combinatorial problem that is receiving much attention from the neural combinatorial community. We propose a generative model based on the well-known Pointer Network and train it with SLIM. Experiments on popular benchmarks demonstrate the potential of this approach as the resulting models outperform constructive heuristics and state-of-the-art learning proposals for the JSP. Lastly, we prove the robustness of SLIM to various parameters and its generality by applying it to the Traveling Salesman Problem.