TransGP: Task-Conditioned Transformer-Guided Genetic Programming for Multitask Dynamic Flexible Job Shop Scheduling

2026-04-04

AI Summary

This work addresses the slow convergence of traditional genetic programming and the difficulty of existing multi-task approaches in generating task-aware heuristic rules for dynamic flexible job shop scheduling. To overcome these limitations, the authors propose TransGP, a hyper-heuristic framework that integrates task-conditioned Transformers with genetic programming. TransGP is the first to embed task-specific information into a Transformer architecture to generate scheduling heuristics tailored to individual dynamic scenarios, while leveraging an elite heuristic distribution to guide population-based search and enable knowledge transfer across tasks. Experimental results demonstrate that TransGP significantly outperforms baseline methods across various dynamic environments, achieving faster convergence, higher solution quality, and greater robustness.

Abstract

Hyper-heuristics have become a popular approach for solving dynamic flexible job shop scheduling (DFJSS) problems. They use gradient-free optimization techniques like Genetic Programming (GP) to evolve non-differentiable heuristics. However, conventional GP methods tend to converge slowly because they rely solely on evolutionary search to find good heuristics. Existing multitask GP methods can solve multiple tasks simultaneously and speed up the search by transferring knowledge across similar tasks. But they mostly exchange heuristic building blocks without truly generating heuristics conditioned on task information. In this paper, we aim to accelerate convergence and enable task-specific heuristic generation by incorporating a task-conditioned Transformer model. The Transformer works in two ways. First, it learns the distribution of elite heuristics, biasing the search toward promising regions of the heuristic space. Second, through conditional generation, it produces heuristics tailored to specific tasks, allowing the model to handle multiple scheduling tasks at once and improving overall optimization efficiency. Based on these ideas, we propose TransGP, a Task-Conditioned Transformer-Guided GP framework. This evolutionary paradigm integrates generative modeling with GP, enabling efficient multitask heuristic learning and knowledge transfer. We evaluate TransGP on a range of DFJSS scenarios. Experimental results show that TransGP consistently outperforms multitask GP baselines, widely used handcrafted heuristics, and the pure Transformer model, achieving faster convergence, superior solution quality, and enhanced robustness.

Problem

• dynamic flexible job shop scheduling
• genetic programming
• multitask optimization
• hyper-heuristics
• task-conditioned heuristic generation

Innovation

• Task-Conditioned Transformer
• Genetic Programming
• Multitask Optimization
• Hyper-heuristics
• Dynamic Flexible Job Shop Scheduling