FROM COIN 2013 · TO MULTI-OBJECTIVE SCHEDULING

COIN Flow-Shop
Research Lab

A reproducible Python laboratory for rebuilding Node-Based COIN, extending it to multi-objective scheduling, and comparing learned permutation models with modern evolutionary algorithms.

Research → Teaching → PracticeOne evaluator and equal objective-evaluation budgets make every comparison auditable. The lab is intended for classroom explanation, reproducible experiments, presentations, and exploration toward practical scheduling systems.

01 · Research lineage

Revisiting Node-Based COIN

Inspired by Srimongkolkul and Chongstitvatana (JCSSE 2013), which applied Node-Based Coincidence learning to flow-shop scheduling.

Srimongkolkul & Chongstitvatana, 2013 ↗

02 · What this lab adds

Python, reusable models, multiple objectives

The legacy ideas are reconstructed as reusable Python components, extended from scalar fitness to Pareto solution sets, and benchmarked against GA, NSGA-II, SPEA2, NSGA-III, EHBSA and NHBSA.

03 · Why it matters

Explain trade-offs before automating decisions

Use the lab to teach algorithmic learning, present transparent evidence, and prototype scheduling decisions involving completion time, flow time, tardiness and machine utilization before adapting them to business constraints.

StatusReady

Algorithm × seed runs0 / 0

Automatic evaluations100

Best observed · all objectives

—

Instance matrix

Best objective versus evaluations

Pareto / objective comparison

Selected solution schedule

Algorithm · seed · solution

Final solution sets

Each row is one solution. Multi-objective runs return a final Pareto set; single-objective runs return one best solution.

Run an experiment to populate results.