This issue of AIEDAM is based on a workshop on Machine Learning in Design held at the 1994 Conference on Artificial Intelligence in Design, AID'94, (Gero & Sudweeks, 1994); the second of such workshops, with the first being held at AID'92 in 1992 (Gero, 1992). The first workshop also resulted in a special issue of AIEDAM (Maher et al., 1994).

In this paper the need for Intelligent Computer Aided Design (Int.CAD) to jointly support design and learning assistance is introduced. The paper focuses on presenting and exploring the possibility of realizing “learning” assistance in Int.CAD by introducing a new concept called Shared Learning. Shared Learning is proposed to empower CAD tools with more useful learning capabilities than that currently available and thereby provide a stronger interaction of learning between a designer and a computer. “Controlled” computational learning is proposed as a means whereby the Shared Learning concept can be realized. The viability of this new concept is explored by using a system called PERSPECT. PERSPECT is a preliminary numerical design tool aimed at supporting the effective utilization of numerical experiential knowledge in design. After a detailed discussion of PERSPECT's numerical design support, the paper presents the results of an evaluation that focuses on PERSPECT's implementation of “controlled” computational learning and ability to support a designer's need to learn. The paper then discusses PERSPECT's potential as a tool for supporting the Shared Learning concept by explaining how a designer and PERSPECT can jointly learn. There is still much work to be done before the full potential of Shared Learning can be realized. However, the authors do believe that the concept of Shared Learning may hold the key to truly empowering learning in Int.CAD.

New methods of configuration analysis have recently emerged that are based on development trends characteristic of many technical systems. It has been found that though the development of any system aims to increase a combination of the performance, reliability and economy, actual design changes are frequently kept to a minimum to reduce the risk of failure. However, a strategy of risk reduction commits the designer to an existing configuration and an approved set of components and materials. Therefore, it is important to analyze the configurations, components, and materials of past designs so that good aspects may be reused and poor ones changed. A good configuration produces the required performance and reliability with maximum economy. These three evaluation criteria form the core of a configuration optimization tool called KATE, where known configurations are optimized producing a set of ranked trial solutions. The authors suggest that this solution set contains valuable design knowledge that can be reused. This paper briefly introduces a generic method of configuration evaluation and then describes the use of a self-organizing neural network, the Kohonen Feature Map, to analyze solution sets by performing an initial data reduction step, producing archetype solutions, and supporting qualitative clustering.

This paper describes current research toward automating the redesign process. In redesign, a working design is altered to meet new problem specifications. This process is complicated by interactions between different parts of the design, and many researchers have addressed these issues. An overview is given of a large design tool under development, the Circuit Designer's Apprentice. This tool integrates various techniques for reengineering existing circuits so that they meet new circuit requirements. The primary focus of the paper is one particular technique being used to reengineer circuits when they cannot be transformed to meet the new problem requirements. In these cases, a design plan is automatically generated for the circuit, and then replayed to solve all or part of the new problem. This technique is based upon the derivational analogy approach to design reuse. Derivational Analogy is a machine learning algorithm in which a design plan is saved at the time of design so that it can be replayed on a new design problem. Because design plans were not saved for the circuits available to the Circuit Designer's Apprentice, an algorithm was developed that automatically reconstructs a design plan for any circuit. This algorithm, Reconstructive Derivational Analogy, is described in detail, including a quantitative analysis of the implementation of this algorithm.

Learning engineering is a new subarea of knowledge engineering dealing with the methodological aspects of using learning systems in knowledge acquisition. In this paper, the justification for the development of Learning Engineering is provided, and its major subdomains and research directions are briefly discussed.

Analogical reasoning plays an important role in design. In particular, cross-domain analogies appear to be important in innovative and creative design. However, making cross-domain analogies is hard and often requires abstractions common to the source and target domains. Recent work in case-based design suggests that generic mechanisms are one type of abstractions useful in adapting past designs. However, one important yet unexplored issue is where these generic mechanisms come from. We hypothesize that they are acquired incrementally from design experiences in familiar domains by abstraction over patterns of regularity. Three important issues in abstraction from experiences are what to abstract from an experience, how far to abstract, and what methods to use. In this short paper, we describe how structure-behavior-function models of designs in a familiar domain provide the content, and together with the problem-solving context in which learning occurs, also provide the constraints for learning generic mechanisms from design experiences. In particular, we describe the model-based learning method with a scenario of learning feedback mechanism.

We are interested in investigating the different types of knowledge and mechanisms that cause design problem-solving to become routine. We use the model of Routine Parametric Design incorporated in the language DSPL (Brown & Chandrasekaran, 1989), where both the parameters to be given values and the design knowledge needed are known in advance.

Over the past decade “design assistance,” that is, where the computer is viewed as an Intelligent Design Assistant (IDA) (MacCallum et al., 1987), has emerged in knowledge based design support and has formed the basic research strategy for the CAD Centre, University of Strathclyde, since the mid-1980s. Within this philosophy, an IDA would act as a colleague to a designer, providing guidance, learning from past design experiences, carrying out semi- and fully-automated tasks, explaining its reasoning and in essence complementing the designer's own natural skills, and thus leaving the ultimate decision-making, control, and responsibility with the designer (Fig. 1).

Science has developed detailed and well-founded theories for analyzing the behavior of artifacts. For example, Boeing was able to correctly verify an entirely new airplane, the Boeing 777, before any prototype was even built. However, there are few theories, and no computer systems, that would allow us to design structures with a similar degree of automation.

Most of machine learning in design has focussed on learning generalizations to predict some future behavior of the system under consideration. Such approaches have been applied primarily to the analysis and synthesis stages of designing. There has been little work done relating to the formulation stage. This paper applies a particular machine learning approach to the improvement of the formal description of the design formulation. It applies an evolutionary technique to the problem reformulation to improve the formulation. This results in both a near optimal problem formulation and an improvement in the solution synthesized from that formulation.

This research originates in the work started several years ago at Worcester Polytechnic Institute dedicated to the investigation, modelling and evaluation of multiagent based design. The main thrust behind our approach is the idea of finding elementary patterns of agent problem-solving and interaction in design tasks. To achieve this goal we introduced and defined the concept of Single Function Agents, (SiFAs) (Dunskus, 1995; SiFA, 1995). SiFAs are agents specialized to perform one single generic function during the design process. Some typical functions would be selection, evaluation, and critique. These types of agents can be instantiated for different, particular design domains.

According to Simon (1983), machine learning is a process that improves the performance of an intelligent system. The performance task of a design system is to arrive at a structural description of a device given its functional description. Machine learning can occur, to improve the performance of the design system, in a number of ways. The learning can make the process of arriving at a known structure to function mapping faster, either by improving the control strategy or by learning new world knowledge needed for the structure to function mapping. It can also improve the performance of the design system by allowing it to come up with new structure to function mappings. The latter kinds of learning leads to innovative designs. The work presented in this paper, EnviroAdapt, belongs to the latter category. EnviroAdapt has the performance task of designing devices along with their operating environments. The EnviroAdapt learns about the novel environments and designs devices for these environments. This learning process is enabled by adapting the previous designs of devices along with their operating environments. This adaptation process makes use of abstracting general mechanisms from the previous designs, then instantiating them in the context of new design problem. Both of these processes are driven by the characteristics of the new design problem that requires a device to operate within a novel operating environment.

SALT is the highly influential system implemented by Marcus (1988), which is able to acquire knowledge and solve problems using the propose-and-revise strategy. The primary task to which this has been applied is the design of elevators, or lifts (Marcus et al., 1988). Recently we have re-implemented and extended the system at Aberdeen, S-SALT, the new system, has the following enhancements:

Most computer-based design tools assume designers work with a well-defined problem. The traditional treatment of design as two discrete phases; problem formulation and solution synthesis, is challenged by recent research. Though the view on discrete phases may be applicable to simple and/or well-defined design tasks, current research (Jonas, 1993; Logan & Smithers, 1993; Gero, 1994; Smithers et al., 1994) has shown that design is an ill-structured problem and the discrete phases view is not a good description of the process during which design alternatives are generated. A potential role of machine learning techniques is to provide a computational model of the changing representation of the design problem in response to the search for design solutions.

In this abstract, we discuss the use of learning techniques to improve performance and solution quality in multiagent parametric design. We have implemented the L-TEAM testbed for empirical evaluation of two forms of learning (described in detail below):

Though simulation models are extensively used for detailed design analysis, they find limited role in preliminary design decisions. We have developed a machine learning based approach to enable detailed simulation models to be harvested for supporting early-stage design of engineering systems.

Since the inception of research on machine learning (ML), these techniques have been associated with the task of automated knowledge generation or knowledge reorganization. This association still prevails, as seen in this issue. When the use of ML programs began to attract researchers in engineering design, different existing tools were used to test their utility and gradually, variations of these tools and methods have sprung up. In many cases, the use of these tools was based on availability and not necessarily applicability. When we began working on ML in design, we attempted to follow a different path (Reich, 1991a; Reich & Fenves, 1992) that led to the design of Bridger (Reich & Fenves, 1995), a system for learning bridge synthesis knowledge. Subsequent experiences and further reflection led us to conclude that the process of using ML in design requires careful and systematic treatment for identifying appropriate ML programs for executing the learning tasks we wish to perform (Reich, 1991b, 1993a). Another observation was that the task of creating or reorganizing knowledge for real design tasks is outside the scope of present ML programs. Establishing the practical importance of ML techniques had to start by addressing engineering problems that could benefit from present ML programs.

Creative design has been characterized in computational terms as “that design activity which occurs when a new variable is introduced into the design” (Gero, 1994). This is opposed to “routine design,” where “knowledge about variables, objectives expressed in terms of those variables, constraints expressed in terms of those variables, and the processes needed to find values for those variables, are all known a priori.” A third alternative, “innovative design,” occurs when no new variables are introduced, but when one or more variables are used with values outside the usual scope. In computational terms, routine design can also be seen as “search,” (sometimes also called “exploitation”) in so far as a certain, predefined search space is searched for a design solution. For creative design, where the search proceeds outside the boundaries of a predefined search space, the term “exploration” can be used.

We are working on using machine learning to make the numerical optimization of complex engineering designs faster and more reliable. We envision a system that learns from previous design sessions knowledge that enables it to assist the engineer in setting up and carrying out a new design optimization. We have performed initial experiments for two aspects of setting up an optimization: selecting a prototype to serve as a starting point for the optimization and selecting a reformulation of the search space. Both choices can dramatically affect the speed and the reliability of design optimization.