Linguistic Fuzzy Modeling for High Dimensional Regression Problem Using Multi-Objective Genetic Algorithm
Sayeeda Ashraf, Pintu Chandra Shill · 2018 International Conference on Computer, Communication, Chemical, Material and Electronic Engineering (IC4ME2) · 2018
Fuzzy logic controllers suffer from the curse of dimensionality problem since the number of rules in a standard fuzzy system increases exponentially with the number of input and output variables. One way to overcome this problem is to decide the utilized linguistic variables, partitioning the linguistic variable and the rule base together, in order to only evolve very simple, but still accurate models. In order to accomplish these, we propose a novel 2-tuple linguistic fuzzy model and linguistic fuzzy partitioning technique. Our propose 2-tuple linguistic fuzzy logic controller tackles the curse of dimensionality problem in high-dimensional regression problems when the number of input and output variables becomes high. The linguistic information can be expressed by means of 2-tuples (S,α), where S is the linguistic term and α is the numeric value between [-0.5, 0.5]. Here, tackle 2-tuple fuzzy linguistic model capable for making processes of computing with words (CW) without loss of information. In order to validate our proposed method, we solve the high dimensional regression problem: length and maintenance cost estimation of low and medium voltage line respectively. We present extensive simulation results in order to demonstrate the simplicity and superiority of the proposed technique while comparing with other methods.