Machine learning in evolutionary economics
Zhen Ye · 2001
Evolutionary economics is a promising method to build artificial intelligent systems to solve difficult tasks. An improved method utilizing reinforcement learning is proposed here with a Block World problems experiment presented to explain the building of an evolutionary market. The experimental system evolved from empty to a universal algorithm in only 66000 training instances, to solve tasks for any number of blocks with a success rate of 0.92. Other merits of the GA algorithm are also discussed.