Title :
An evolutionary approach to shape emergence
Author :
Yan, Min ; Dai, Ruwei
Author_Institution :
Inst. of Autom., Chinese Acad. of Sci., Beijing, China
Abstract :
Shape emergence is a recognized visual phenomenon experienced by visually all humans. It involves perception of emergent shapes that only implicitly exist in a primary shape (an already interpreted shape), and do not directly correspond to the entities or subshapes used to construct the primary shape. This paper presents an evolutionary approach to shape emergence. Its basic idea is to maintain a population of shapes that aggregate with each other to produce their offspring shapes. The shape population evolves from one generation to another as new shapes are produced and old shapes are eliminated. The evolution is governed by a fitness function and a set of local aggregating rules. Since offspring shapes are always larger in size than their parent shapes, the evolution finally stops with a quiescent population in which no shape can produce any further offspring because of the limitation of the global boundary of the primary shape. The last several generations of shapes in the evolving process provide a vocabulary for emergent interpretations of the primary shape
Keywords :
genetic algorithms; inference mechanisms; pattern recognition; visual perception; emergent interpretations; emergent shapes; evolutionary; evolutionary approach; offspring shapes; population of shapes; primary shape; shape emergence; visual phenomenon; Aggregates; Automation; Humans; Lattices; Pattern recognition; Problem-solving; Psychology; Shape; Vocabulary;
Conference_Titel :
Evolutionary Computation, 1994. IEEE World Congress on Computational Intelligence., Proceedings of the First IEEE Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1899-4
DOI :
10.1109/ICEC.1994.349958