Robust online active learning

In many industrial applications, obtaining labeled observations is not straightforward as it often requires the intervention of human experts or use expensive testing equipment. these circumstances, active learning can be highly beneficial in suggesting most informative data points to used when fitting a model. Reducing number needed for model development alleviates both computational burden required training and operational expenses related labeling. Online learning, particular, useful high-volume production processes where decision about acquisition label point needs taken within an extremely short time frame. However, despite recent efforts develop online strategies, behavior methods presence outliers has been thoroughly examined. this work, we investigate performance linear regression contaminated streams. Our study shows that currently available query strategies are prone sample outliers, whose inclusion set eventually degrades predictive models. To address issue, propose solution bounds search area conditional D-optimal algorithm uses robust estimator. approach strikes balance between exploring unseen regions input space protecting against outliers. Through numerical simulations, show proposed method effective improving thus expanding potential applications powerful tool.