The stripe-backed weasel Mustela strigidorsa: taxonomy, ecology, distribution and status

1. The stripe-backed weasel Mustela strigidorsa is one of the rarest and least-known mustelids in the world. Its phylogenetic relationships with other Mustela species remain controversial, though several unique morphological features distinguish it from congeners. 2. It probably lives mainly in evergreen forests in hills and mountains, but has also been recorded from plains forest, dense scrub, secondary forest, grassland and farmland. Known sites range in altitude from 90 m to 2500 m. Data are insufficient to distinguish between habitat and altitudes which support populations, and those where only dispersing animals may occur. 3. It has been confirmed from many localities in north-east India, north and central Myanmar, south China, north Thailand, north and central Laos, and north and central Vietnam. Given the limited survey effort, the number of recent records shows that the species is not as rare as hitherto believed. Neither specific nor urgent conservation needs are apparent.

Smoothing LDA Model for Text Categorization

Abstract. Latent Dirichlet Allocation (LDA) is a document level language model. In general, LDA employ the symmetry Dirichlet distribution as prior of the topic-words’ distributions to implement model smoothing. In this paper, we propose a data-driven smoothing strategy in which probability mass is allocated from smoothing-data to latent variables by the intrinsic inference procedure of LDA. In such a way, the arbitrariness of choosing latent variables'priors for the multi-level graphical model is overcome. Following this data-driven strategy,two concrete methods, Laplacian smoothing and Jelinek-Mercer smoothing, are employed to LDA model. Evaluations on different text categorization collections show data-driven smoothing can significantly improve the performance in balanced and unbalanced corpora.