Pu WangPhD Student (June 2008 - )
Nature Inspired Computationa and Applications Laboratory (NICAL)
University of Science and Technology of China
Background: EDDIE for ForecastingJin Li used constraints to direct Genetic Programming (GP) in its search for forecasting strategies. This allows his program to trade between "rate of failure" and "rate of missing chances" (*Note). In Jin Li's GP, which is called FGP, constraint takes a form of window, defined by two numbers [Min, Max]. The user has to input these constraints to FGP. Performance of FGP is sometimes sensitive to the constraints used. The values of [Min, Max] are often sensitive to the stock prices.
Research Direction: Constraint-directed GPIn this research, we propose to make the constraints part of the evolution. Instead of providing [Min, Max], users will supply GP with constraints which are directly related to forecasting needs, such as "maximum rate of failure", "maximum rate of missing chances", or a combination of them. GP should adjust [Min, Max] to satisfy these constraints. Additional or alternative constraints should be explored to adjust GP's behaviour in order to satisfy the users' constraints.
Supervisors:Pu Wang is a student at Nature Inspired Computationa and Applications Laboratory (NICAL), University of Science and Technology of China, where he is jointly supervised by Professor Xin Yao (Director of NICAL) and Visiting Professor Edward Tsang, assisted by Dr Ke Tang at USTC.
Maintained by Edward Tsang; updated 2 July 2008