Dimensions of Machine Learning in Design
{Version: Fri Feb 14 21:26:16 EST 1997}
- What can trigger learning?
- Failure; Success; Differences between expected and real values
- Need to improve abilities
- What are the elements supporting learning?
- Critique, praise, estimates, evaluations and advice (internal or external)
- Sequences of design decisions
- Design histories, e.g traces of information flow, knowledge exchange, negotiation
- Analyses of failures and conflicting elements (goals, decisions)
- Feedback after completing the design task
- What might be learned?
- Constraints relating parameters or other elements of the design
- Dependencies between design parameters
- Support in favor of or against a decision
- Design rules, methods and plans
- Analogical associations
- Preferences
- Preconditions and postconditions for rules, actions and tasks
- Consequences of design decisions
- Types of failures and conflicts
- Heuristics for failure recovery and conflict resolution
- Successful designs and design processes
- Availability of knowledge for learning
- Direct communication (with the user or another design system)
- Indirect communication (e.g., between design systems via a blackboard)
- Record of the state of the design
- Repositories of design and interaction histories
- Methods of learning
- Explanation-based learning
- Induction
- Knowledge compilation
- Case-based and analogical learning
- Reinforcement learning
- Genetic algorithms
- Neural networks
- Local vs. Global Learning
- Learning by a design program
- Learning by a group of design programs
- Consequences of learning
- Design improvement
- Improvement of the design process
Adapted from Grecu & Brown,
Dimensions of Learning in Agent-based Design,
AID'96, June 1996.