Data labelling and ground truth
Assigning correct answers to training examples.
Supervised models learn from labels, so label errors and inconsistent labelling directly harm performance. Controls include clear labelling guidelines, multiple labellers, agreement checks and quality sampling. Ground truth is the reference answer that model outputs are compared against.
On the exam
Check how labels were created and quality-checked. Biased labels produce biased models.
Related terms
- Supervised, unsupervised and reinforcement learningThree main ways models learn.
- Data quality for AIWhether data is fit to train and run a model.
AAIA Test yourself on this with timed AAIA practice questions in the AAIA Exam Simulation Hub, or read the free AAIA exam guide.
Test yourself with today’s question of the day.
