Improving Data Mining Via Noisy Micro-Outsourcing

This work discussed in this paper focuses on problems where it is possible to obtain certain (noisy) data values ("Labels") relatively cheaply, from on-line micro outsourcing sources ("Non-expert labelers"). A main focus is the strategy of outsourcing to obtain these values as training labels for supervised modeling. (This setting is in direct contrast to the setting motivating active learning and semi-supervised learning, where unlabeled points are relatively inexpensive, but labeling is expensive.)