ImitAL: Learned Active Learning Strategy on Synthetic Data
This work addresses the challenge of developing robust active learning strategies for researchers and practitioners across various domains, though it is incremental as it builds on existing heuristic combinations.
The paper tackles the problem of inconsistent performance across datasets in active learning query strategies by proposing ImitAL, a domain-independent method that learns an optimal combination of informativeness and representativeness heuristics through training on synthetic data, achieving competitive results in evaluations on 13 diverse datasets.
Active Learning (AL) is a well-known standard method for efficiently obtaining annotated data by first labeling the samples that contain the most information based on a query strategy. In the past, a large variety of such query strategies has been proposed, with each generation of new strategies increasing the runtime and adding more complexity. However, to the best of our our knowledge, none of these strategies excels consistently over a large number of datasets from different application domains. Basically, most of the the existing AL strategies are a combination of the two simple heuristics informativeness and representativeness, and the big differences lie in the combination of the often conflicting heuristics. Within this paper, we propose ImitAL, a domain-independent novel query strategy, which encodes AL as a learning-to-rank problem and learns an optimal combination between both heuristics. We train ImitAL on large-scale simulated AL runs on purely synthetic datasets. To show that ImitAL was successfully trained, we perform an extensive evaluation comparing our strategy on 13 different datasets, from a wide range of domains, with 7 other query strategies.