Marine Litter Windrows: A Strategic Target to Understand and Manage the Ocean Plastic Pollution

Windrow is a long-established term for the aggregations of seafoam, seaweeds, plankton and natural debris that appear on the ocean surface. Here, we define a “litter windrow” as any aggregation of floating litter at the submesoscale domain (<10 km horizontally), regardless of the force inducing the surface convergence, be it wind or other forces such as tides or density-driven currents. The marine litter windrows observed to date usually form stripes from tens up to thousands of meters long, with litter densities often exceeding 10 small items (<2 cm) per m2 or 1 large item (>2 cm) per 10 m2. Litter windrows are generally overlooked in research due to their dispersion, small size and ephemeral nature. However, applied research on windrows offers unique possibilities to advance on the knowledge and management of marine litter pollution. Litter windrows are hot spots of interaction with marine life. In addition, since the formation of dense litter windrows requires especially high loads of floating litter in the environment, their detection from space-borne sensors, aerial surveys or other platforms might be used to flag areas and periods of severe pollution. Monitoring and assessing of management plans, identification of pollution sources, or impact prevention are identified as some of the most promising fields of application for the marine litter windrows. In the present Perspective, we develop a conceptual framework and point out the main obstacles, opportunities and methodological approaches to address the study of litter windrows.

Frequent Patterns Mining

Frequent pattern mining has been an important subject matter in data mining from many years. A remarkable progress in this field has been made and lots of efficient algorithms have been designed to search frequent patterns in a transactional database. One of the most important technique of datamining is the extraction rule in large database. The time required for generating frequent itemsets plays an important role. This paper provides a comparative study of algorithms Eclat, Apriori and FP-Growth. The performance of these algorithms is compared according to the efficiency of the time and memory usage. This study also focuses on each of the algorithm’s strengths and weaknesses for finding patterns among large item sets in database systems.

Gradient-Based Optimization for Bayesian Preference Elicitation

Effective techniques for eliciting user preferences have taken on added importance as recommender systems (RSs) become increasingly interactive and conversational. A common and conceptually appealing Bayesian criterion for selecting queries is expected value of information (EVOI). Unfortunately, it is computationally prohibitive to construct queries with maximum EVOI in RSs with large item spaces. We tackle this issue by introducing a continuous formulation of EVOI as a differentiable network that can be optimized using gradient methods available in modern machine learning computational frameworks (e.g., TensorFlow, PyTorch). We exploit this to develop a novel Monte Carlo method for EVOI optimization, which is much more scalable for large item spaces than methods requiring explicit enumeration of items. While we emphasize the use of this approach for pairwise (or k-wise) comparisons of items, we also demonstrate how our method can be adapted to queries involving subsets of item attributes or “partial items,” which are often more cognitively manageable for users. Experiments show that our gradient-based EVOI technique achieves state-of-the-art performance across several domains while scaling to large item spaces.

Application of Using Simulated Annealing to Combine Clustering with Collaborative Filtering for Item Recommendation

tem-item collaborative filtering was widely used in item recommender system because of good recommend effects. However when facing a large amount of items, there would be performance reduction, because of building a very large item comparison dataset in order to find the similar item. K-means cluster had a very good effect in classification and a good performance even though the dataset being processed is very large. But the cold start was a problem to k-means and we must do some extra work to use it in item recommendation. By using the simulated annealing theory to combine the two methods to fixed the problems of the two methods mentioned above and take use of their advantages for better recommendation effect and performance. The experimental results show that, using simulated annealing to combine the clustering and collaborative filtering in item recommendation system can get more stable recommendation results of better quality.

Ecological Factors and Diversification among Neotropical Characiforms

Morphological and DNA sequence data has been used to propose hypotheses of relationships within the Characiformes with minimal comparative discussion of causes underpinning the major intraordinal diversification patterns. We explore potential primary morphological factors controlling the early diversification process in some Neotropical characiforms as the first step to identifying factors contributing to the pronounced intraordinal morphological and species diversity. A phylogenetic reconstruction based on 16S rDNA (mitochondrial) and 18S rDNA (nuclear) genes provided the framework for the identification of the main morphological differences among the Acestrorhynchidae, Anostomidae, Characidae, Ctenoluciidae, Curimatidae, Cynodontidae, Gasteropelecidae, Prochilodontidae and Serrasalmidae. Results indicate an initial split into two major groupings: (i) species with long dorsal-fin bases relative to the size of other fins (Curimatidae, Prochilodontidae, Anostomidae, Serrasalmidae) which primarily inhabit lakes, swamps, and rivers (lineage I); and (ii) species with short dorsal-fin bases (Acestrorhynchidae, Gasteropelecidae, Characidae) which primarily inhabit creeks and streams (lineage II). The second diversification stage in lineage I involved substantial morphological diversification associated with trophic niche differences among the monophyletic families which range from detritivores to large item predators. Nonmonophyly of the Characidae complicated within lineage II analyzes but yielded groupings based on differences in pectoral and anal fin sizes correlated with life style differences.