Hybrid Deep Neural Network Scheduler for Job-Shop Problem Based on Convolution Two-Dimensional Transformation

Table 10

Comparison tables with traditional classification methods on ZLP datasets.

		Near optimal	HDNNS (our)	ANN (1D)	KNN	SVM	DT	ERT	GOSS
ZLP (15  12); pU (10, 40)	Average MAKESPAN	320.14	334.88	369.68	424.03	426.29	403.20	398.69	400.18
	Scheduling score	—	95.6	86.6	75.5	75.1	79.4	80.3	80
ZLP (15  12); pU (10, 50)	Average MAKESPAN	375.04	398.55	434.57	498.72	512.34	482.67	464.73	472.34
	Scheduling score	—	94.1	86.3	75.2	73.2	77.7	80.7	79.4
ZLP (15  12); pU (10, 60)	Average MAKESPAN	443.97	450.27	513.85	584.94	599.96	581.11	547.44	550.15
	Scheduling score	—	98.6	86.4	75.9	74	76.4	81.1	80.7
ZLP (15  12); pU (10, 70)	Average MAKESPAN	513.13	551.16	591.16	676.95	686.00	645.45	636.64	635.85
	Scheduling score	—	93.1	86.8	75.8	74.8	79.5	80.6	80.7
ZLP (15  12); pU (10, 80)	Average MAKESPAN	560.94	618.46	653.78	751.93	744.94	705.59	702.93	704.70
	Scheduling score	—	90.7	85.8	74.6	75.3	79.5	79.8	79.6
ZLP (15  18); pU (10, 40)	Average MAKESPAN	476.40	508.43	550.75	617.90	649.05	606.88	597.74	599.25
	Scheduling score	—	93.7	86.5	77.1	73.4	78.5	79.7	79.5
ZLP (15  18); pU (10, 50)	Average MAKESPAN	571.62	617.97	660.83	737.57	774.55	725.41	711.86	732.85
	Scheduling score	—	92.5	86.5	77.5	73.8	78.8	80.3	78
ZLP (15  18); pU (10, 60)	Average MAKESPAN	681.60	745.73	789.80	898.02	926.09	874.97	846.71	858.44
	Scheduling score	—	91.4	86.3	75.9	73.6	77.9	80.5	79.4
ZLP (15  18); pU (10, 70)	Average MAKESPAN	794.26	849.48	920.35	1034.19	1074.78	1024.85	990.35	1019.59
	Scheduling score	—	93.5	86.3	76.8	73.9	77.5	80.2	77.9
ZLP (15  18); pU (10, 80)	Average MAKESPAN	900.40	971.31	1043.34	1176.99	1210.22	1151.41	1138.31	1151.41
	Scheduling score	—	92.7	86.3	76.5	74.4	78.2	79.1	78.2
Average	Average MAKESPAN	563.75	604.62	652.81	740.13	760.42	720.15	703.54	712.47
	Scheduling score	—	93.59	86.38	76.08	74.15	78.34	80.23	79.34
Average MAKESPAN rank			1	2	6	7	5	3	4