An In-depth Benchmarking of Evolutionary and Swarm Intelligence Algorithms for Autoscaling Parameter Sweep Applications on Public Clouds

Table 13

Average hypervolume of the Pareto sets provided by each algorithm (first autoscaling stage) for each of the applications and sizes. Higher values represent better values.

Application	Size	NSGA-III	SMPSO	SMS-EMOA	E-NSGA-III
MPG	30	67	26	43	74
	100	84	27	23	94
	300	33	40	12	54
GMS	30	54	51	38	70
	100	49	28	42	62
	300	70	48	44	81
FPA	30	94	48	56	96
	100	83	22	71	87
	300	76	38	29	81
WRF	30	17	14	18	29
	100	24	13	22	34
	300	18	14	16	31

The significance level of bold values given in Table 13 is α = 0.001 (significance level).