|
Contact |
![]() |
Hamid
R. Tizhoosh Faculty of Engineering University of Waterloo ![]() Pattern Recognition, Computer Vision, Medical Imaging, Machine Intelligence |
![]() |
Affiliations …
|
OBL Research Group focuses on potentials of concept of opposition in machine intelligence. The group consists of several members each working on a different aspect of OBL. Opposition-based learning (OBL) is a new scheme for machine learning,
search and optimization, which uses opposite values, counter-estimates,
opposite weights, anti-chromosomes and opposites states and actions
in order to accelerate existing methodologies such as neural networks,
reinforcement learning, evolutionary algorithms and swarm-based
methods.
|
|
The underlying ideal of OBL is quite simple: Whenever we are guessing we should look at the opposite guess as well. By doing this, either the guess or the opposite guess possesses lower error, is fitter, or receives more reward; The learning, search or optimization will then continue with the better one, either with guess or with opposite guess.
|
Created
by:
Log
Web Design