TY - GEN
T1 - Enhanced Dyna-QPC model with Fuzzy logic to train gaming models
AU - Ignatious, Henry Alexander
AU - Hesham-El-Sayed,
AU - Khan, Manzoor Ahmad
N1 - Funding Information:
ACKNOWLEDGMENT This paper is supported by Emirates Center for Mobility Research of the United Arab Emirates University (grant 31R271)
Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - This paper presents an automated learning process to train the mountain car game model. It proposes an Enhanced Dyna-QPC model to effectively train the mountain car model in the stipulated time, based on their perceived environmental conditions. Decision Tree (DT) classification model along with Neural Network (NN)) model is used in this research to frame decision rules and self-train the game model respectively. Discrete Finite Deterministic Automata (DFA) concepts are included to finalize the state transition of the training model. Moreover, the Erdos-Renyi Random graph-generating model is used to generate dynamic state transition graphs to minimize the number of states. To increase the range of conditions and to derive meaningful decision rules, fuzzy concepts are used in this paper. Various simulation experiments have been conducted to evaluate the efficiency of the proposed training process. Simulation results reveal better performance over 3 popular models in the literature.
AB - This paper presents an automated learning process to train the mountain car game model. It proposes an Enhanced Dyna-QPC model to effectively train the mountain car model in the stipulated time, based on their perceived environmental conditions. Decision Tree (DT) classification model along with Neural Network (NN)) model is used in this research to frame decision rules and self-train the game model respectively. Discrete Finite Deterministic Automata (DFA) concepts are included to finalize the state transition of the training model. Moreover, the Erdos-Renyi Random graph-generating model is used to generate dynamic state transition graphs to minimize the number of states. To increase the range of conditions and to derive meaningful decision rules, fuzzy concepts are used in this paper. Various simulation experiments have been conducted to evaluate the efficiency of the proposed training process. Simulation results reveal better performance over 3 popular models in the literature.
KW - Decision Tree Classification Model
KW - Dyna-QPC
KW - Finite Deterministic Automata
KW - Fuzzy Logic and Erdos-Renyi
KW - Neural Network Model
UR - http://www.scopus.com/inward/record.url?scp=85126744523&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85126744523&partnerID=8YFLogxK
U2 - 10.1109/GCAIoT53516.2021.9692963
DO - 10.1109/GCAIoT53516.2021.9692963
M3 - Conference contribution
AN - SCOPUS:85126744523
T3 - 2021 IEEE Global Conference on Artificial Intelligence and Internet of Things, GCAIoT 2021
SP - 25
EP - 30
BT - 2021 IEEE Global Conference on Artificial Intelligence and Internet of Things, GCAIoT 2021
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2021 IEEE Global Conference on Artificial Intelligence and Internet of Things, GCAIoT 2021
Y2 - 12 December 2021 through 16 December 2021
ER -