Artificial neural network optimization with Levenberg-Maruardt algorithm for dynamic gesture recognition

• Main Authors: Dy, Stephen John, Gonzales, Matthew Adrianne, Lozano, Lenard, Suniga, Miguel Angelo, Abad, Alexander C.
• Format: text
• Published: Animo Repository 2018
• Subjects: Robots, Industrial; Neural networks (Computer science); Electrical and Computer Engineering; Electrical and Electronics
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/3375
• Institution: De La Salle University

Movement has long been a mode of expression and communication. A challenge arises when we try to bestow the ability to learn and recognize movements to machines, specifically computers, but with the development of sensor technology and the growing interest in machine learning algorithms, there is an opportunity to explore and formulate new approaches. The study focuses on the use of the Levenberg Marquardt Algorithm as an optimization algorithm for a multilayer Artificial Neural Network in constructing a predictive model for dynamic gestures. Extraction of the data set was made integral to the research. The study concludes that the network architecture is adequate for gesture recognition, with an average recognition rate of 83%, but a larger data set may show to improve this value. © 2018 Authors.