DeepAI AI Chat
Log In Sign Up

Active Learning Helps Pretrained Models Learn the Intended Task

by   Alex Tamkin, et al.
Stanford University

Models can fail in unpredictable ways during deployment due to task ambiguity, when multiple behaviors are consistent with the provided training data. An example is an object classifier trained on red squares and blue circles: when encountering blue squares, the intended behavior is undefined. We investigate whether pretrained models are better active learners, capable of disambiguating between the possible tasks a user may be trying to specify. Intriguingly, we find that better active learning is an emergent property of the pretraining process: pretrained models require up to 5 times fewer labels when using uncertainty-based active learning, while non-pretrained models see no or even negative benefit. We find these gains come from an ability to select examples with attributes that disambiguate the intended behavior, such as rare product categories or atypical backgrounds. These attributes are far more linearly separable in pretrained model's representation spaces vs non-pretrained models, suggesting a possible mechanism for this behavior.


page 2

page 5

page 6

page 7

page 14

page 15

page 16


Learning a Policy for Opportunistic Active Learning

Active learning identifies data points to label that are expected to be ...

Active Learning of Non-semantic Speech Tasks with Pretrained Models

Pretraining neural networks with massive unlabeled datasets has become p...

Bayesian Active Learning with Pretrained Language Models

Active Learning (AL) is a method to iteratively select data for annotati...

ActiveLab: Active Learning with Re-Labeling by Multiple Annotators

In real-world data labeling applications, annotators often provide imper...

Can I see an Example? Active Learning the Long Tail of Attributes and Relations

There has been significant progress in creating machine learning models ...

Improving OCR Accuracy on Early Printed Books by combining Pretraining, Voting, and Active Learning

We combine three methods which significantly improve the OCR accuracy of...

An Active Approach for Model Interpretation

Model interpretation, or explanation of a machine learning classifier, a...