Special Issue: Machine learning in design

Alex H.B. Duffy, David C. Brown, Mary Lou Maher
1996 Artificial intelligence for engineering design, analysis and manufacturing  
The linking of research in machine learning with research in knowledge-based design is such that each of the two areas benefit from the consideration of the other. The use of machine learning in design addresses the perceived need to support the capture and representation of design knowledge, because handcrafting a representation is a difficult and time-consuming task. In addition, design provides a task with which to investigate the usefulness of existing machine learning techniques, and,
more » ... ps, to discover new ones. Knowledge-based design systems are typically encoded as compiled experience, in the form of generalized rules and/or objects. Machine learning research provides a set of theories and methods for learning from a variety of sources, such as from uncompiled experience in the form of examples of previous problems. Thus, the design knowledge need not be completely handcrafted. The knowledge learned can be used to solve problems faster, or better, or to solve problems that could not be solved before. From the perspective of design, knowledge-based design systems provide support for a wide range of activities broadly decomposed into two tasks, synthesis and evaluation/criticism. Synthesis systems are concerned with producing a design solution or solutions from a set of specifications. Evaluation systems are concerned with determining performance or critiquing a given design. The form of knowledge needed for a synthesis system differs from that needed for an evaluation system, although both types of knowledge can be induced from design experience. The role of machine learning, and the methods which are appropriate, varies according to the type of knowledge being learned.
doi:10.1017/s0890060400001323 fatcat:i3dtydztqjcnhnyoe5ry4ejogy