WORD · brick

machine teaching

Machine teaching is machine learning run backwards: instead of finding the concept from the examples, you know the concept and compute the smallest set of examples that will plant it in a particular learner.

The learner can be a program or a person; either way the teacher needs a model of how that learner learns, and the lesson is only as good as the model. It matters here because it makes distillation — shrinking knowledge to fit a mind — a computable objective rather than a craft (machine-distillation).

Links

← back to the gate