flowshop_permutation_problem

Graph-Based Imitation Learning for Permutation Flow Shop Scheduling (PFSS)

📌 Project Overview

This is an implementation of the paper Learning to Optimize Permutation Flow Shop Scheduling via Graph-based Imitation Learning (check it via this link https://arxiv.org/abs/2210.17178 ) presents a novel approach to solving the Permutation Flow Shop Scheduling (PFSS) problem, a complex optimization challenge commonly found in manufacturing and production systems. The model learns from expert solutions (NEH heuristic) and uses a Graph Neural Network (GNN) with attention mechanisms
to predict optimal job sequences that minimize the makespan.

🚀 Features

• Implements Graph-Based Imitation Learning (IL) using Gated Graph Convolutional Networks (GGCN).
• Learns scheduling from expert solutions (NEH heuristic).
• Trains using supervised learning and optimizes using CrossEntropyLoss.
• Supports variable job and machine configurations.
• Provides evaluation results with predicted sequences and makespan.