Search

The bluespotted cornetfish (Fistularia commersonii) (Osteichtyes, Fistulariidae) is considered to be one of the most invasive species of the Mediterranean Sea and Europe but only scattered information exists on its distribution and abundance. Here we collated the available species records, following its first detection in the Mediterranean Sea, in January 2000, until October 2011. A total of 191 observations were used to reconstruct the invasion sequence, to provide estimates of the rate of spread and to construct an environmental suitability model based on six biophysical variables and the maximum entropy approach. The results showed that colonization of the Mediterranean Sea proceeded in parallel along the southern and northern rim of the Basin at speeds that reached 1,000–1,500 km year−1 with a clear decrease in the rate of spread at the Sicily Strait. The most important explanatory variables for describing the distribution of F. commersonii were mean depth (explaining 32.4 % of the data variance), chlorophyll-a (29.3 %), and salinity (18.4 %). Coastal areas with relatively low chlorophyll-a concentrations and high salinity were the preferred habitat of the bluespotted cornetfish in its invaded range. Conversely, extreme productivity (highly eutrophic or highly oligotrophic), low salinity and cold temperatures provided abiotic resistance to this invasion. Areas of high environmental suitability were identified along the northern coasts of the Levantine Sea, Dodecanese, Sicily Strait and Tyrrhenian Sea. In contrast, the north Aegean Sea, the Adriatic and the Alboran Sea, the Nile Delta, the western coasts of Egypt and Cyrenaica were unfavourable for the invasion. Despite some limits due to the model’s resolution scale, these general predictions provide new insights into the F. commersonii invasion, indicating abiotic factors of primary importance in shaping the distribution of this species in its invaded range

Data sparseness or overfitting is a serious problem in natural language processing employing machine learning methods. This is still true even for the maximum entropy (ME) method, whose flexible modeling capability has alleviated data sparseness more successfully than the other probabilistic models in many NLP tasks. Although we usually estimate the model so that it completely satisfies the equality constraints on feature expectations with the ME method, complete satisfaction leads to undesirable overfitting, especially for sparse features, since the constraints derived from a limited amount of training data are always uncertain. To control overfitting in ME estimation, we propose the use of box-type inequality constraints, where equality can be violated up to certain predefined levels that reflect this uncertainty. The derived models, inequality ME models, in effect have regularized estimation with L_1 norm penalties of bounded parameters. Most importantly, this regularized estimation enables the model parameters to become sparse. This can be thought of as automatic feature selection, which is expected to improve generalization performance further. We evaluate the inequality ME models on text categorization datasets, and demonstrate their advantages over standard ME estimation, similarly motivated Gaussian MAP estimation of ME models, and support vector machines (SVMs), which are one of the state-of-the-art methods for text categorization., identifier:08856125, identifier:https://dspace.jaist.ac.jp/dspace/handle/10119/3305