Adaptive Bi-objective Genetic Programming for Data-Driven System Modeling

Abstract

We propose in this paper a modification of one of the modern state-of-the-art genetic programming algorithms used for data-driven modeling, namely the Bi-objective Genetic Programming (BioGP). The original method is based on a concurrent minimization of both the training error and complexity of multiple candidate models encoded as Genetic Programming trees. Also, BioGP is empowered by a predator-prey co-evolutionary model where virtual predators are used to suppress solutions (preys) characterized by a poor trade-off error vs complexity. In this work, we incorporate in the original BioGP an adaptive mechanism that automatically tunes the mutation rate, based on a characterization of the current population (in terms of entropy) and on the information that can be extracted from it. We show through numerical experiments on two different datasets from the energy domain that the proposed method, named BioAGP (where "A" stands for "Adaptive"), performs better than the original BioGP, allowing the search to maintain a good diversity level in the population, without affecting the convergence rate.

Smart and intelligent energy monitoring systems: A comprehensive literature survey and future research guidelines

“…For instance, house buildings with small surface and roof areas will require little amount of energy to obtain optimal heating and cooling temperature. For instance, in, 85 Tran Nguyen et al [101] Rasel et al [102] Do et al [103] Zhang et al [104] Li et al [105] Adams et al [106] Electricity Load Diagrams [67] Amri et al [107] Moslemi et al [108] Jin et al [109] --Premarathne et al [110] Hosseini et al [111] Ertugrul et al [31] Moslemi et al [108] Sen et al [113] Fenza et al [114] Bicego et al [115] Chen et al [116] SML2010 [120] Bevilacqua et al [118] Zamora-Martínez et al [117] Maggiolo et al [120] --Qin et al [121] Wollsen et al [122] Li et al [123] Liu et al [124] Combined Cycle Power Plant Dataset [126] Janoušek et al [126] Rashid et al [127] --Sun et al [128] Roni et al [129] Elfaki et al [130] Cao et al [131] 2006 and November 2010. There are some missing parameters (about 1.25%) in the dataset that are represented by the absence of values between two consecutive semi-colon attribute separators.…”