The Viability of B. tuberculosis (Bovinus) on Pasture Land, In Stored Faeces and In Liquid Manure

The object of these experiments was to determine the length of time during which B. tuberculosis in cow's faeces remain alive and virulent on pasture land in the south of England. The method of testing for living B. tuberculosis is given in Appendix II.

Demography and population viability of polar bears in the Gulf of Boothia, Nunavut

We estimated demographic parameters and harvest risks for polar bears (Ursus maritimus) inhabiting the Gulf of Boothia, Nunavut, from 1976 to 2000. We computed survival and abundance from capture–recapture and recovery data (630 marks) using a Burnham joint live–dead model implemented in program MARK. Annual mean total survival (including harvest) was 0.889 ± 0.179 ( x ± 1 SE) for cubs, 0.883 ± 0.087 for subadults (ages 1–4), 0.919 ± 0.044 for adult females, and 0.917 ± 0.041 for adult males. Abundance in the last 3 yr of study was 1,592 ± 361 bears. Mean size of newborn litters was 1.648 ± 0.098 cubs. By age 7, 0.97 ± 0.30 of available females were producing litters. Harvest averaged 38.4 ± 4.2 bears/year in the last 5 yr of study; however, the 2002–2007 kill averaged 56.4 bears/yr. We used a harvested Population Viability Analysis (PVA) to examine impacts of increasing rates of harvest. We estimated the current population growth rate, λH, to be 1.025 ± 0.032. Although this suggests the population is growing, progressive environmental changes may require more frequent population inventory studies to maintain the same levels of harvest risk.

An Extended Kernel for Generalized Multiple-Instance Learning

The multiple-instance learning (MIL) model has been successful in areas such as drug discovery and content-based image-retrieval. Recently, this model was generalized and a corresponding kernel was introduced to learn generalized MIL concepts with a support vector machine. While this kernel enjoyed empirical success, it has limitations in its representation. We extend this kernel by enriching its representation and empirically evaluate our new kernel on data from content-based image retrieval, biological sequence analysis, and drug discovery. We found that our new kernel generalized noticeably better than the old one in content-based image retrieval and biological sequence analysis and was slightly better or even with the old kernel in the other applications, showing that an SVM using this kernel does not overfit despite its richer representation.