Explanation-based learning
term_id: explanation_based_learning
Category: training_techniques
Definition
EBL combines symbolic reasoning with machine learning to accelerate the learning process. Instead of relying on large datasets, it takes a single positive example and uses a pre-existing domain theory to explain why the example belongs to the target concept. This explanation is then operationalized into a general rule that can be applied to future instances. It is particularly useful when data is scarce but domain knowledge is abundant, allowing for rapid acquisition of skills.
Summary
Explanation-based learning is a machine learning method that uses domain theory and a single training example to generate an explanation, which is then generalized into a new rule.
Key Concepts
- Domain Theory
- Operationalization
- Single Example Learning
- Symbolic Reasoning
Use Cases
- Robotics skill acquisition with limited examples
- Automated theorem proving
- Diagnostic systems in specialized technical fields