Dual Strategy Active Learning

By Pinar E. Donmez, Jaime G. Carbonell, Paul N. Bennett. 2007

224 citations recorded by the source corpus.

Active Learning methods rely on static strategies for sampling unlabeled point(s). These strategies range from uncertainty sampling and density estimation to multi-factor methods with learn-once-use-always model parameters. This paper proposes a dynamic approach, called DUAL, where the strategy selection parameters are adaptively updated based on estimated future residual error reduction after each actively sampled point. The objective of dual is to outperform static strategies over a large operating range: from very few to very many labeled points. Empirical results over six datasets demonstrate that DUAL outperforms several state-of-the-art methods on most datasets.

Subfields: Artificial Intelligence, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

Citations in the corpus

Efficient Meta Lifelong-Learning with Limited Memory (2020, methodology): “We consider two popular paradigms in active learning (Donmez et al., 2007), namely the diversity-based method that picks the most representative examples and the uncertainty-based method that picks the most unsure examples.”
Cost-Effective Incremental Deep Model: Matching Model Capacity With the Least Sampling (2023, unknown): “E.g., Donmez et al. proposed a dual strategy for active learning that exploits both measures [43], Huang et al. utilized model capacity degree to actively adapt a pre-trained model with less labeled examples [44].”
A theory of transfer learning with applications to active learning (2012, background): “Active learning is a powerful form of supervised machine learning characterized by interaction between the learning algorithm and data source during the learning process (Cohn et al. 1994; McCallum and Nigam 1998; Campbell et al. 2000; Tong and Koller 2001; Nguyen and Smeulders 2004; Baram et al. 20”
Adaptive Active Learning for Image Classification (2013, methodology): “The methods in [6, 20] explicitly combine clustering and active learning together to exploit both labeled and unlabeled instances.”