节点文献
“不可预知”的设计教学——人工智能思路的数字化设计教学初探
"Unpredictable" Design Teaching——A Premliminary Study on the Digital Design Teaching of AI
【Author】 Liu Xiaokai;Zhang Fan;Fan Wenbing;Zhao Dongmei;Wang Haoyu;Department of Architecture,School of Design,Shanghai Jiaotong University;
【机构】 上海交通大学设计学院建筑系;
【摘要】 计算机技术的本质是用简单的规则实现复杂的功能,就像图片识别最终阅读的是矩阵像素点一样,最简单的规则往往就是复杂形态的开始。在算法上,元胞自动机与人工智能的学习思路很相似,用简单的规则创造了复杂并且完全无法预知的结果,这个思路启发了本课程的作业设置。本文通过一次弱化人的主观形态控制的设计教学过程,仅制定一个简单的规则来形成最后的建筑形态,有意识地将人工智能思维运用于建筑设计过程,并尝试在这个过程中加入机器监督学习的方法。在建筑设计的人工智能技术还未成熟的阶段,探讨人工智能思维对于教学的启发。
【Abstract】 The essence of computer technology is to implement complex functions with simple rules.Just as image recognition finally leads to pixels of matrix,the simplest rules are often the beginning of complex forms.In terms of algorithm, the cellular automata is very similar to the way of the artificial intelligence thinking,and the simple rules create complex and completely unpredictable results.This idea inspired this course.Through a design teaching process that weakens the subjective form control of human beings,this paper only formulates a simple rule to form the final architectural form,consciously applies artificial intelligence thinking to the architectural design process,and attempts to join the machine supervision learning process in this process.In the stage of artificial intelligence technology in architecture,the inspiration of artificial intelligence thinking for teaching is discussed.
【Key words】 Artificial Intelligence; Algorithm; Machine Learning; Supervised Learning; Cellular Automata;
- 【会议录名称】 2019中国高等学校建筑教育学术研讨会论文集
- 【会议名称】2019全国高等学校建筑教育学术研讨会暨院长系主任大会(2019教育部高等学校建筑学专业教学指导分委会年会)
- 【会议时间】2019-10-18
- 【会议地点】中国四川成都
- 【分类号】G642;TU201-4
- 【主办单位】教育部高等学校建筑学专业教学指导分委会、西南交通大学