Towards optimal selection schedules for online active learning

Key points

• We investigate the optimal query policy, that prescribes whether and when a sample should be labeled and used for a training step, or instead discarded, so as to minimize a given objective.1
• We focus on the realizable case, under Gaussian design, in the high-dimensional regime where data dimension and number of unlabeled samples grow proportionally.1
• Taking inspiration from the myopic policy for the cumulative cost, we design a very simple query policy schedule that we compare to other proposed heuristics, on synthetic data and real medical image datasets.1
• Finally, we discuss myopic policies for two-layer neural networks, and exemplify how naive margin-based policies can cease to be optimal.1

Sources (1)

• [1]Towards optimal selection schedules for online active learning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 01:54 PM

  We consider the problem of online linear classification, where a stream of unlabeled samples is available, and annotation is costly.

  We investigate the optimal query policy, that prescribes whether and when a sample should be labeled and used for a training step, or instead discarded, so as to minimize a given objective.

Extractive summary: sentences quoted from the sources.