Page 53 - CyberSecurityEssentialsEnglish
P. 53
Term Common abbreviation Plain-language explanation
Predicting a number or continuous
Regression - value; called numerical forecasting in
this chapter.
Performing well on training examples
Overfitting -
but poorly on new ones.
The ability to succeed on new cases
Generalization -
that were not seen during training.
Examples used during development to
Validation set - compare versions and settings before
final evaluation.
Examples held back from training and
Test set -
used for final evaluation.
A change in the data or operating
Dataset shift - environment that reduces
performance after deployment.
A system that works with more than
Multimodal system - one type of input, such as text, images,
audio, and files.
Accepting an automated output
Automation bias - because it is presented confidently,
without adequate review.
A model trained broadly and then
adapted to many tasks, used as a
Foundation model -
general-purpose starting point rather
than built for one job.
A tuning stage in which people compare
candidate answers and the model is
Learning from
- adjusted toward the ones they
human feedback
preferred; it shapes helpfulness, not
truth.
52
