Can Active Learning Experience Be Transferred?

TL;DR

This paper explores whether active learning strategies can be transferred across datasets by learning and updating strategy weights using a bandit approach, demonstrating improved performance and transferability.

Contribution

It introduces a novel model that linearly combines active learning strategies and employs a bandit algorithm to transfer experience across datasets.

Findings

Transferred experience improves active learning performance.

The model is competitive with existing strategies on single datasets.

Experience transfer enhances future learning tasks.

Abstract

Active learning is an important machine learning problem in reducing the human labeling effort. Current active learning strategies are designed from human knowledge, and are applied on each dataset in an immutable manner. In other words, experience about the usefulness of strategies cannot be updated and transferred to improve active learning on other datasets. This paper initiates a pioneering study on whether active learning experience can be transferred. We first propose a novel active learning model that linearly aggregates existing strategies. The linear weights can then be used to represent the active learning experience. We equip the model with the popular linear upper- confidence-bound (LinUCB) algorithm for contextual bandit to update the weights. Finally, we extend our model to transfer the experience across datasets with the technique of biased regularization. Empirical studies demonstrate that the learned experience not only is competitive with existing strategies on most single datasets, but also can be transferred across datasets to improve the performance on future learning tasks.