The western gray whale: a review of past exploitation, current status and potential threats

Gray whales (Eschrichtius robustus) occur along the eastern and western coastlines of the North Pacific as two geographically isolated populations and have traditionally been divided into the eastern (California-Chukchi) and western (Korean-Okhotsk) populations. Recent molecular comparisons confirm, based on differences in haplotypic frequencies, that these populations are genetically separated at the population-level. Both populations were commercially hunted, but only the eastern gray whale has returned to near pre-exploitation numbers. In contrast, the western population remains highly depleted, shows no apparent signs of recovery and its future survival remains uncertain. Research off Sakhalin Island, Russia between 1995 and 1999 has produced important new information on the present day conservation status of western gray whales and provided the basis for the World Conservation Union (IUCN) to list the population as 'Critically Endangered in 2000. The information presented here, in combination with potential impacts from anthropogenic threats throughout the range of this population, raises strong concerns about the recovery and continued survival of the western gray whale.

Balancing Exploration and Exploitation: A New Algorithm for Active Machine Learning

Active machine learning algorithms are used when large numbers of unlabeled examples are available and getting labels for them is costly (e.g. requiring consulting a human expert). Many conventional active learning algorithms focus on refining the decision boundary, at the expense of exploring new regions that the current hypothesis misclassifies. We propose a new active learning algorithm that balances such exploration with refining of the decision boundary by dynamically adjusting the probability to explore at each step. Our experimental results demonstrate improved performance on data sets that require extensive exploration while remaining competitive on data sets that do not. Our algorithm also shows significant tolerance of noise.