New active learning strategy (disagreement based high influential instances)

[Rafik Bouguelia]

Hello,

For those of you who are interested in new active learning query strategies, here is a recent paper presenting an interesting method to measure the importance of samples to be queried:

M. R. Bouguelia, et al. (2017). Agreeing to disagree: active learning with noisy labels without crowdsourcing. International Journal of Machine Learning and Cybernetics, 1-13.

URL:  https://link.springer.com/article/10.1007/s13042-017-0645-0

Abstract:
We propose a new active learning method for classification, which handles label noise without relying on multiple oracles (i.e., crowdsourcing). We propose a strategy that selects (for labeling) instances with a high influence on the learned model. An instance x is said to have a high influence on the model h, if training h on x (with label y=h(x)) would result in a model that greatly disagrees with h on labeling other instances. Then, we propose another strategy that selects (for labeling) instances that are highly influenced by changes in the learned model. An instance x is said to be highly influenced, if training h with a set of instances would result in a committee of models that agree on a common label for x but disagree with h(x). We compare the two strategies and we show, on different publicly available datasets, that selecting instances according to the first strategy while eliminating noisy labels according to the second strategy, greatly improves the accuracy compared to several benchmarking methods, even when a significant amount of instances are mislabeled.

Best regards,
Mohamed-Rafik BOUGUELIA

[zhouxu1...@gmail.com]

Nice work！