English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In active learning, sampling bias could pose a serious inconsistency problemand hinder the algorithm from finding the optimal hypothesis. However, manymethods for neural networks are hypothesis space agnostic and do not addressthis problem. We examine active learning with convolutional neural networksthrough the principled lens of version space reduction. We identify theconnection between two approaches---prior mass reduction and diameterreduction---and propose a new diameter-based querying method---the minimumGibbs-vote disagreement. By estimating version space diameter and bias, weillustrate how version space of neural networks evolves and examine therealizability assumption. With experiments on MNIST, Fashion-MNIST, SVHN andSTL-10 datasets, we demonstrate that diameter reduction methods reduce theversion space more effectively and perform better than prior mass reduction andother baselines, and that the Gibbs vote disagreement is on par with the bestquery method.

Effective Version Space Reduction for Convolutional Neural Networks