Exploitation and Exploration

John Holland's principle of balancing exploitation and exploration (Holland, 1992): the right balance between exploitation and exploration allows a complex adaptive system to focus on the most promising possibilities seen so far, while looking for new possibilities at the same time as well. Pure exploitation would not allow the acquisition of new information, pure exploration would not allow to use available resources efficiently. Examples are according to Melanie Mitchell the immune system or an ant colony which search for intruders and food, respectively: when promising possibilities are identified, they should be exploited at a rate and intensity related to their estimated promise, which is continually updated. But at all times exploration for new possibilities should continue.

Therefore the principle is important for adaptive information processing in changing environments. It is also important in many other fields, for example in reinforcement learning and economics: should companies invest in research & development and fund innovative ideas or should they try to use their available resources and techniques more efficiently ?

The principle of balancing exploitation and exploration is based on two core principles of evolution: exploration means variation (for example by recombination or mutation), while natural selection is similar to exploitation.

Books

• Holland, J. H. 1992. Adaptation in Natural and Artificial Systems. Cambridge, MA: MIT Press. First edition, 1975