Evaluation

What is active learning?

Active learning chooses which unlabeled examples a person should label next, so the model improves faster than if you labeled at random.

Updated

How it works

  • Score unlabeled rows by uncertainty or disagreement.
  • Send the top of that list to a person.
  • Write the new labels back onto the rows and repeat.

What it is not

It is not labeling every row, and it is not a random sample.

active learning: this, and the thing it is confused with

active learning: this, and the thing it is confused with
ThisNot this
SelectsInformative rowsAll rows, or a random draw
Labels goBack on the rowInto a file nobody joins
LoopModel, rank, labelA single export

Where Pixeltable fits

Pixeltable can rank rows with a computed score and store the human label on the same row, which is the loop.

Questions

How does active learning work?
Score unlabeled rows by uncertainty or disagreement. Send the top of that list to a person. Write the new labels back onto the rows and repeat.
What is active learning often confused with?
It is not labeling every row, and it is not a random sample.