TY - GEN
T1 - Moving towards explainable artificial intelligence using fuzzy rule-based networks in decision-making process
AU - Arabikhan, Farzad
AU - Gegov, Alexander
AU - Taheri, Rahim
AU - Akbari, Negar
AU - Bader-El-Den, Mohamed
N1 - Publisher Copyright:
© 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2024/1/20
Y1 - 2024/1/20
N2 - Advanced Machine Learning and Artificial Intelligence techniques are very pow-erful in predictive tasks and they are getting more popular as decision making tools across many industries and fields. However, they are mostly weak in ex-plaining the inference and internal process and they are referred to as black-box models. Fuzzy Rule Based Network is a powerful white-box technique which maps well the external inputs, intermediate latent variables and outputs a modular approach based on Fuzzy Logic and it is capable of dealing with complexity and linguistic uncertainty in decision making process. To improve the performance of Fuzzy Rule Based Network, it requires to be tuned and optimized to increase its accuracy, transparency and efficiency. In this paper, a method is proposed to tune the Fuzzy Rule Based Network by using Fuzzy C-Mean and Genetic Algorithm for rule reduction and tuning membership functions and also Backward Selection techniques for pruning and input and branch selection. A case study in transport and telecommuting is used to illustrate the performance of the proposed method. The results show the Fuzzy Rule Based Network’s ability to explain the internal process of decision making and its capabilities in transparency, interpretability and in moving towards Explainable Artificial Intelligence (XAI).
AB - Advanced Machine Learning and Artificial Intelligence techniques are very pow-erful in predictive tasks and they are getting more popular as decision making tools across many industries and fields. However, they are mostly weak in ex-plaining the inference and internal process and they are referred to as black-box models. Fuzzy Rule Based Network is a powerful white-box technique which maps well the external inputs, intermediate latent variables and outputs a modular approach based on Fuzzy Logic and it is capable of dealing with complexity and linguistic uncertainty in decision making process. To improve the performance of Fuzzy Rule Based Network, it requires to be tuned and optimized to increase its accuracy, transparency and efficiency. In this paper, a method is proposed to tune the Fuzzy Rule Based Network by using Fuzzy C-Mean and Genetic Algorithm for rule reduction and tuning membership functions and also Backward Selection techniques for pruning and input and branch selection. A case study in transport and telecommuting is used to illustrate the performance of the proposed method. The results show the Fuzzy Rule Based Network’s ability to explain the internal process of decision making and its capabilities in transparency, interpretability and in moving towards Explainable Artificial Intelligence (XAI).
KW - Fuzzy Rule Based Network
KW - Fuzzy Rule Based Network Tuning
KW - Decision Making Process
KW - White-Box Model
KW - Explainable AI
UR - https://iciks.org/
UR - https://iciks.org/paper-submission-iciks2023/
U2 - 10.1007/978-3-031-51664-1_21
DO - 10.1007/978-3-031-51664-1_21
M3 - Conference contribution
AN - SCOPUS:85184281290
SN - 9783031516634
T3 - Lecture Notes in Business Information Processing
SP - 296
EP - 306
BT - Advances in Information Systems, Artificial Intelligence and Knowledge Management
A2 - Saad, Inès
A2 - Rosenthal-Sabroux, Camille
A2 - Gargouri, Faiez
A2 - Chakhar, Salem
A2 - Williams, Nigel
A2 - Haig, Ella
PB - Springer
T2 - 6th International Conference on Information and Knowledge Systems, ICIKS 2023
Y2 - 22 June 2023 through 23 June 2023
ER -