引入均衡探索范式的创意众包进化计算方法

发布时间：2018-05-21 20:05

  本文选题：创意众包 + 进化计算　； 参考：《机械工程学报》2017年15期

【摘要】：创意众包通过招募大众完成产品概念、平面广告等创意设计工作,是获取创意方案的重要手段。但创意众包中参与者通常为非专业设计人员,参与者资源整合不易。为此,提出一种引入均衡探索范式的创意众包进化计算方法,结合设计专家的专业能力与计算机的计算能力整合众包参与者资源。该方法依据设计认知领域中对专家行为的研究成果改进创意众包流程,使众包群体行为更接近专家的设计过程,提升产出的创意方案质量。具体的,利用专家行为的均衡探索范式,包括"持续信息引入"、"结构化探索"、"广度深度均衡探索"、"新老方案均衡探索"等行为准则,计算创意方案的进化适应度,改进进化计算中的基因编码、交叉遗传等操作,引导众包的方案进化。为验证此创意众包方法,实施了一个创意众包实例,招募252名参与者就坐姿矫正产品利用提出的方法产出三代创意方案。实例验证该方法能够扩展众包探索的设计空间,提升众包的创意方案质量。
[Abstract]:Creative crowdsourcing is an important means to obtain creative solutions by recruiting the public to complete the creative design work such as product concept, print advertising and so on. However, participants in creative crowdsourcing are usually non-professional designers, participants in the integration of resources is not easy. In this paper, an evolutionary computing method of creative crowdsourcing based on equilibrium exploration paradigm is proposed, which combines the professional ability of design experts with the computing power of computer to integrate crowdsourcing participant resources. According to the research results of expert behavior in the field of design cognition, this method improves the process of creative crowdsourcing, makes crowdsourcing group behavior closer to the design process of experts, and improves the quality of creative schemes. In particular, the evolutionary adaptation of creative schemes is calculated by using the norms of equilibrium exploration of expert behavior, including "continuous information introduction", "structured exploration", "breadth and depth equilibrium exploration", "balanced exploration of new and old programs", and so on. Improve genetic coding, cross-inheritance and other operations in evolutionary computing to guide the evolution of crowdsourcing schemes. In order to verify this creative crowdsourcing method, an example of creative crowdsourcing was implemented. 252 participants were recruited to produce three generation creative schemes based on the proposed method. Examples show that this method can expand the design space of crowdsourcing and improve the quality of crowdsourcing.
【作者单位】： 浙江大学计算机辅助设计与图形学国家重点实验室;浙江大学现代工业设计研究所;同济大学设计创意学院;
【基金】：国家重点基础研究发展计划(973计划,2015CB352503) 国家自然科学基金(61672451,51675382)资助项目
【分类号】：TB47
，

本文编号：1920587