Expand description
Uncertainty sampling — query the points where a single model is least sure.
Operates on a predicted class-probability matrix (n_samples × n_classes, each row
a distribution); returns per-sample informativeness and a ranking.
Enums§
- Strategy
- Which uncertainty measure to rank by. All are oriented so higher = more informative (more worth a human’s label).
Functions§
- most_
informative - The single most-informative pool index (the next item to ask a human about).
- rank_
informative - Rank pool indices most-informative first under
strategy. - row_
score - Informativeness of one predicted distribution under
strategy. - score
- Per-sample informativeness for a probability matrix (each inner slice a row).