Interactive Evolutionary Computation (IEC) is a derivative class of Evolutionary Computation which incorporates interaction with human users. Most IEC applications fall into a category known as “narrowly defined IEC” (NIEC). In NIEC, the task of fitness evaluation is outsourced to human users. For example, a user may be presented with a visual representation of the current generation of individuals. The user is then prompted to provide fitness information about some or all of the individuals. The computer in turn uses this fitness information to produce the next generation of individuals through the application of a predefined sequence evolutionary operators.
We propose Hyperinteractive Evolutionary Computation (HIEC), a novel form of IEC in which a human user actively chooses when and how to apply each of the available evolutionary operators, playing the central role in the control flow of evolutionary search processes. We present HIEC as a novel form of IEC where a human user has all available evolutionary operators at his/her disposal. The user directs the overall search process and initiates actions by choosing when and how each evolutionary operator is applied. The user may add a new individual to the population through the crossover, mutate, duplicate, or random operators. The user can also remove individuals with the delete operator. This naturally results in dynamic variability of population size and continuous generation change (like steady-state strategies for Genetic Algorithms).
In HIEC, the user can wield evolutionary operators like tools, using each to impart a different kind of specific change to a subset of the evolving population of individuals, just as a painter uses a variety of brushes and paints to impart different kinds of change to the developing canvas. In this sense, working with an HIEC system is somewhat similar to working with typical interactive editing applications.
Moreover, while NIEC is characterized by abrupt changes from one generation to the next, the population in an HIEC changes more gradually, with only a small number of individuals being added or deleted at any given time. While a gradual population changes in the (non-interactive) evolutionary computation literature abound, HIEC is the first example of an IEC framework to use this technique.
There is another important difference between HIEC and NIEC. While NIEC requires that the user explicitly supply fitness information to the system, no such mechanism for obtaining fitness information from the user exists in HIEC. This is because in HIEC it is the human, not the computer, who decides when and how to perform selection.
It sounds like Hyperinteractive Evolutionary Computation (HIEC) is a form of Interactive Evolutionary Computation (IEC) in which a human user has greater control over the evolutionary search process. In HIEC, the user can choose when and how to apply various evolutionary operators, such as crossover, mutation, duplication, and deletion. As a result, the population changes more gradually, with only a small number of individuals being added or deleted at any given time. This is in contrast to "narrowly defined IEC" (NIEC), in which the user provides fitness information to the system and the computer applies a predefined sequence of evolutionary operators to produce the next generation of individuals. HIEC is also distinguished from NIEC by the fact that it does not require the user to explicitly supply fitness information to the system, as the user is responsible for selecting which individuals will be included in the population. Is there anything else I can help with?