A Survey of Underwater Acoustic Data Classification Methods Using Deep Learning for Shoreline Surveillance.

This paper presents a comprehensive overview of current deep-learning methods for automatic object classification of underwater sonar data for shoreline surveillance, concentrating mostly on the classification of vessels from passive sonar data and the identification of objects of interest from active sonar (such as minelike objects, human figures or debris of wrecked ships). Not only is the contribution of this work to provide a systematic description of the state of the art of this field, but also to identify five main ingredients in its current development: the application of deep-learning methods using convolutional layers alone; deep-learning methods that apply biologically inspired feature-extraction filters as a preprocessing step; classification of data from frequency and time-frequency analysis; methods using machine learning to extract features from original signals; and transfer learning methods. This paper also describes some of the most important datasets cited in the literature and discusses data-augmentation techniques. The latter are used for coping with the scarcity of annotated sonar datasets from real maritime missions.

Genomic prediction in contrast to a genome-wide association study in explaining heritable variation of complex growth traits in breeding populations of Eucalyptus.

BACKGROUND: The advent of high-throughput genotyping technologies coupled to genomic prediction methods established a new paradigm to integrate genomics and breeding. We carried out whole-genome prediction and contrasted it to a genome-wide association study (GWAS) for growth traits in breeding populations of Eucalyptus benthamii (n =505) and Eucalyptus pellita (n =732). Both species are of increasing commercial interest for the development of germplasm adapted to environmental stresses. RESULTS: Predictive ability reached 0.16 in E. benthamii and 0.44 in E. pellita for diameter growth. Predictive abilities using either Genomic BLUP or different Bayesian methods were similar, suggesting that growth adequately fits the infinitesimal model. Genomic prediction models using ~5000-10,000 SNPs provided predictive abilities equivalent to using all 13,787 and 19,506 SNPs genotyped in the E. benthamii and E. pellita populations, respectively. No difference was detected in predictive ability when different sets of SNPs were utilized, based on position (equidistantly genome-wide, inside genes, linkage disequilibrium pruned or on single chromosomes), as long as the total number of SNPs used was above ~5000. Predictive abilities obtained by removing relatedness between training and validation sets fell near zero for E. benthamii and were halved for E. pellita. These results corroborate the current view that relatedness is the main driver of genomic prediction, although some short-range historical linkage disequilibrium (LD) was likely captured for E. pellita. A GWAS identified only one significant association for volume growth in E. pellita, illustrating the fact that while genome-wide regression is able to account for large proportions of the heritability, very little or none of it is captured into significant associations using GWAS in breeding populations of the size evaluated in this study. CONCLUSIONS: This study provides further experimental data supporting positive prospects of using genome-wide data to capture large proportions of trait heritability and predict growth traits in trees with accuracies equal or better than those attainable by phenotypic selection. Additionally, our results document the superiority of the whole-genome regression approach in accounting for large proportions of the heritability of complex traits such as growth in contrast to the limited value of the local GWAS approach toward breeding applications in forest trees.

Intake and ingestive behavior in lambs fed low-digestibility forages.

Ingestive behavior of lambs fed diets consisting of fresh sugarcane with urea, bagasse treated with calcium oxide, and urea ammoniated sugarcane bagasse supplemented with concentrate mixture in 50:50 ratio were evaluated. For this, 34 wethers Santa Inês in their growing phase, with an average age of 3.0 ± 0.6 months and a mean initial live weight of 17.8± 5.2 kg were used. The animals were distributed in a completely randomized design and subjected to visual observation periods of 5 days, for 24 h a day, during the experimental period. Dry matter (DM) intake and intake efficiency of DM were higher (P < 0.05) for animals receiving fresh sugarcane with urea. The animals which were fed with bagasse treated with calcium oxide had higher (P < 0.05) consumption of neutral detergent fiber, longer feeding and chewing time (P < 0.05), and shorter (P < 0.05) idling time. The time spent on chewing the ruminal bolus did not differ from one diet to the other (P > 0.05). Grams of dry matter per ruminated bolus were similar among animals fed with fresh sugarcane and ammoniated bagasse (P > 0.05) but lower (P < 0.05) in animals fed with bagasse treated with calcium oxide. Rumination efficiency values, in grams of dry matter per hour, and grams of neutral detergent fiber per hour for all three diets were similar (P > 0.05) to those found for feeding efficiency. The number of feeding and rumination periods was not affected (P > 0.05) by diet. Based on the intake and ingestive behavior responses, the fresh sugarcane with urea compared to bagasse treated with calcium oxide and ammoniated bagasse was found to be the better alternative feed for use in lamb diets.

Exploring the knowledge contained in neuroimages: statistical discriminant analysis and automatic segmentation of the most significant changes.

OBJECTIVE: The aim of this article is to propose an integrated framework for extracting and describing patterns of disorders from medical images using a combination of linear discriminant analysis and active contour models. METHODS: A multivariate statistical methodology was first used to identify the most discriminating hyperplane separating two groups of images (from healthy controls and patients with schizophrenia) contained in the input data. After this, the present work makes explicit the differences found by the multivariate statistical method by subtracting the discriminant models of controls and patients, weighted by the pooled variance between the two groups. A variational level-set technique was used to segment clusters of these differences. We obtain a label of each anatomical change using the Talairach atlas. RESULTS: In this work all the data was analysed simultaneously rather than assuming a priori regions of interest. As a consequence of this, by using active contour models, we were able to obtain regions of interest that were emergent from the data. The results were evaluated using, as gold standard, well-known facts about the neuroanatomical changes related to schizophrenia. Most of the items in the gold standard was covered in our result set. CONCLUSIONS: We argue that such investigation provides a suitable framework for characterising the high complexity of magnetic resonance images in schizophrenia as the results obtained indicate a high sensitivity rate with respect to the gold standard.