Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 39064 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Vosughi Vahdat, Bijan
- Abstract:
- Recently, the computational algorithms underlying in the nervous system of vertebrates have been attracting scientists and engineers. Therefore, a number of artificial learning methods are proposed for mathematical interpretation of these algorithms. In this thesis, by investigating the latest findings about the nervous system, and the motor system in particular, a novel machine learning scheme inspired by human motor learning is proposed. Basic theoretical aspects of this method in conjunction with some of state-of-theart artificial learning methods are discussed. Finally, this method is evaluated in a variety of engineering problems ranging from curve fitting and function approximation to chaotic time series prediction and learning the inverse dynamic of 2-joint arm robot. Simulation results show that this one-pass online learning method, without having any training phase, has a high computational performance and speed and outperforms the other renowned learning schemes. Plausibile compability of this method with the neural mechanisms of the human nervous system, not only bestows a novel insight to neuroscientists about the functionality of the brain, but also provides a small introduction to infant technologies of neuro-robotics and Brain-Machine Interfaces
- Keywords:
- Artificial Intelligence ; Motor Learning ; Motion System ; Human Purposive Movement ; Function Appromination
Digital Object List
-
محتواي پايان نامه
- view
Bookmark