TY  - JOUR
T1  - GHS + LEM: Global-best Harmony Search using learnable evolution models
JO  - Applied Mathematics and Computation
VL  - 218
IS  - 6
SP  - 2558
EP  - 2578
PY  - 2011/11//
T2  - 
AU  - Cobos, Carlos
AU  - Estupiñán, Dario
AU  - Pérez, José
SN  - 0096-3003
M3  - doi: 10.1016/j.amc.2011.07.073
UR  - http://www.sciencedirect.com/science/article/pii/S0096300311010228
KW  - Harmony Search
KW  - Meta-heuristics
KW  - Evolutionary algorithms
KW  - Optimization
KW  - Learnable evolution models
KW  - Machine learning
KW  - Prism
AB  - 
This paper presents a new optimization algorithm called GHS + LEM, which is based on the Global-best Harmony Search algorithm (GHS) and techniques from the learnable evolution models (LEM) to improve convergence and accuracy of the algorithm. The performance of the algorithm is evaluated with fifteen optimization functions commonly used by the optimization community. In addition, the results obtained are compared against the original Harmony Search algorithm, the Improved Harmony Search algorithm and the Global-best Harmony Search algorithm. The assessment shows that the proposed algorithm (GHS + LEM) improves the accuracy of the results obtained in relation to the other options, producing better results in most situations, but more specifically in problems with high dimensionality, where it offers a faster convergence with fewer iterations.
ER  -