We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Efficacy of Bayesian Neural Networks in Active Learning

Abstract: Obtaining labeled data for machine learning tasks can be prohibitively expensive. Active learning mitigates this issue by exploring the unlabeled data space and prioritizing the selection of data that can best improve the model performance. A common approach to active learning is to pick a small sample of data for which the model is most uncertain. In this paper, we explore the efficacy of Bayesian neural networks for active learning, which naturally models uncertainty by learning distribution over the weights of neural networks. By performing a comprehensive set of experiments, we show that Bayesian neural networks are more efficient than ensemble based techniques in capturing uncertainty. Our findings also reveal some key drawbacks of the ensemble techniques, which was recently shown to be more effective than Monte Carlo dropouts.
Comments: Published at CVPR Workshop on Learning From Limited or Imperfect Data (LLID) 2021
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Cite as: arXiv:2104.00896 [cs.LG]
  (or arXiv:2104.00896v2 [cs.LG] for this version)

Submission history

From: Vineeth Rakesh [view email]
[v1] Fri, 2 Apr 2021 06:02:11 GMT (7470kb,D)
[v2] Mon, 19 Apr 2021 22:19:53 GMT (7471kb,D)

Link back to: arXiv, form interface, contact.