Uniformity-Comprehensive Multiobjective Optimization Evolutionary Algorithm Based on Machine Learning

Table 7

The running time comparison results between MOEA-UCML and three traditional multiobjective optimization algorithms on the DTLZ problem set.

Problem	HypE	MOEAD	MOEADDE	MOEA-UCML
DTLZ1	5.7786e + 1 (2.28e + 1)	2.5043e + 0 (2.73e − 1)	2.1339e + 0 (1.72e − 1)	2.2106e + 0 (8.67e + 0)
DTLZ2	3.4190e + 2 (2.00e + 1)	2.5045e + 0 (2.44e − 1)	2.1360e + 0 (1.52e − 1)	1.7138e + 0 (1.01e + 0)
DTLZ3	1.7855e + 1 (1.96e + 0)	2.5450e + 0 (2.05e − 1)	2.1945e + 0 (1.59e − 1)	1.6234e + 0 (1.16e + 0)
DTLZ4	2.7614e + 2 (4.10e + 1)	2.5628e + 0 (2.26e − 1)	2.1925e + 0 (1.43e − 1)	1.6232e + 0 (5.71e − 1)
DTLZ5	2.2814e + 2 (1.73e + 1)	2.5606e + 0 (2.58e − 1)	2.2153e + 0 (1.20e − 1)	1.5807e + 0 (7.11e − 1)
DTLZ6	4.2194e + 2 (3.51e + 1)	2.7896e + 0 (3.49e − 1)	2.2638e + 0 (1.59e − 1)	1.5644e + 0 (9.17e − 1)
DTLZ7	9.6128e + 1 (1.33e + 1)	2.5970e + 0 (2.32e − 1)	2.2420e + 0 (1.95e − 1)	1.5553e + 0 (8.01e − 1)
	7/0/0	6/0/1	6/0/1

The bold values indicate the optimal performance results of all algorithms for each test function.