Human-based computation: Difference between revisions

Revision as of 23:58, 10 March 2006

Revision as of 22:53, 16 March 2006

Line 1:

In computer science, '''human-based computation''' is a technique when a computational process employs humans to perform its function, using human abilities to solve a certain problem or a set of problems. This approach explores differences in abilities and alternative costs between humans and computer agents to achieve symbiotic human-computer interaction.

In traditional computation, a human employs computer to solve a problem: a human provides a formalized problem description to a computer, and receives a solution to interpret. In human-based computation, the roles are reversed: computer asks a person or often a large number of people to solve a problem, then collects, interprets, and integrates their solutions. Precursors of this idea are interactive programs requesting input from a user, e.g. asking a confirmation to delete a file. However, this concept in its explicit form appeared at the intersection of computer graphics and evolutionary computation. [Sims 91] used human visual perception and esthetic ability to implement evaluation function in evolutionary programming application and evolve pieces of graphic art this way. The crucial difference here is agency: Sim's program was no longer an agent of its user, but rather a coordinator of many human evaluators who became agents of the program. Human-based genetic algorithm (HBGA) is a logical extension and a quite general model of this approach based on the idea of outsourcing. Thus, in HBGA humans also can contribute their innovative solutions into the process and have more control over the functions they are performing. Most implementations of HBGA also have some kind of motivation system that encourages humans to participate.

Line 17:

</table>

</center>

==Methods of human-based computation==

* [[Interactive genetic algorithm]] [Sims 91] IGA enables the user to create an abstract drawing only by selecting his/her favorite images displayed on the computer screen, so human performs fitness computation. [Unemi 1998] Simulated breeding style introduces no explicit fitness, just selection, which is easier for human. In a typical IGA system, a computer program generate an image and presents it to a human user for evaluation/classification.

Line 33 ⟶ 30:

==References==

# [Sims 91] Sims, K.: Artificial Evolution for Computer Graphics, Computer Graphics, 25(4) (SIGGRAPH'91), 319-328 (1991)

# [Unemi 98] Unemi, T.: A Design of multi-field user interface for simulated breeding, Proceedings of the Third Asian Fuzzy and Intelligent System Symposium, 489-494 (1998)