Results from the April 2009 Gulf of Alaska Line Transect Survey (GOALS) in the Navy Training Exercise Area

Little is known about the present-day occurrence of cetaceans found in offshore waters in the Gulf of Alaska; however, whaling records and a few recent surveys have shown this area to be important habitat. The U.S. Navy maintains a maritime training area in the central Gulf of Alaska, east of Kodiak Island, and has requested additional information on marine mammal presence and use of this area. To describe the occurrence and distribution of marine mammals in and around the U.S. Navy training area, a line transect visual and acoustic survey was conducted 10-20 April 2009 from the NOAA ship Oscar Dyson. The primary survey area encompassed nearshore and offshore pelagic waters of the central Gulf of Alaska. Survey lines were designed to provide equal coverage of the nearshore and offshore habitat.

Active Learning to Maximize Area Under the ROC Curve

In active learning, a machine learning algorithmis given an unlabeled set of examples U, and is allowed to request labels for a relatively small subset of U to use for training. The goal is then to judiciously choose which examples in U to have labeled in order to optimize some performance criterion, e.g. classification accuracy. We study how active learning affects AUC. We examine two existing algorithms from the literature and present our own active learning algorithms designed to maximize the AUC of the hypothesis. One of our algorithms was consistently the top performer, and Closest Sampling from the literature often came in second behind it. When good posterior probability estimates were available, our heuristics were by far the best.