Use of Social Information in Seabirds: Compass Rafts Indicate the Heading of Food Patches

Ward and Zahavi suggested in 1973 that colonies could serve as information centres, through a transfer of information on the location of food resources between unrelated individuals (Information Centre Hypothesis). Using GPS tracking and observations on group movements, we studied the search strategy and information transfer in two of the most colonial seabirds, Guanay cormorants (Phalacrocorax bougainvillii) and Peruvian boobies (Sula variegata). Both species breed together and feed on the same prey. They do return to the same feeding zone from one trip to the next indicating high unpredictability in the location of food resources. We found that the Guanay cormorants use social information to select their bearing when departing the colony. They form a raft at the sea surface whose position is continuously adjusted to the bearing of the largest returning columns of cormorants. As such, the raft serves as a compass signal that gives an indication on the location of the food patches. Conversely, Peruvian boobies rely mainly on personal information based on memory to take heading at departure. They search for food patches solitarily or in small groups through network foraging by detecting the white plumage of congeners visible at long distance. Our results show that information transfer does occur and we propose a new mechanism of information transfer based on the use of rafts off colonies. The use of rafts for information transfer may be common in central place foraging colonial seabirds that exploit short lasting and/or unpredictably distributed food patches. Over the past decades Guanay cormorants have declined ten times whereas Peruvian boobies have remained relatively stable. We suggest that the decline of the cormorants could be related to reduced social information opportunities and that social behaviour and search strategies have the potential to play an important role in the population dynamics of colonial animals.     

Could specialization to cold‐water upwelling systems influence gene flow and population differentiation in marine organisms? A case study using the blue‐footed booby,Sula nebouxii

Aim We assessed population differentiation and gene flow across the range of the blue‐footed booby (Sula nebouxii) (1) to test the generality of the hypothesis that tropical seabirds exhibit higher levels of population genetic differentiation than their northern temperate counterparts, and (2) to determine if specialization to cold‐water upwelling systems increases dispersal, and thus gene flow, in blue‐footed boobies compared with other tropical sulids. Location Work was carried out on islands in the eastern tropical Pacific Ocean from Mexico to northern Peru. Methods We collected samples from 173 juvenile blue‐footed boobies from nine colonies spanning their breeding distribution and used molecular markers (540 base pairs of the mitochondrial control region and seven microsatellite loci) to estimate population genetic differentiation and gene flow. Our analyses included classic population genetic estimation of pairwise population differentiation, population growth, isolation by distance, associations between haplotypes and geographic locations, and analysis of molecular variance, as well as Bayesian analyses of gene flow and population differentiation. We compared our results with those for other tropical seabirds that are not specialized to cold‐water upwellings, including brown (Sula leucogaster), red‐footed (S. sula) and masked (S. dactylatra) boobies. Results Blue‐footed boobies exhibited weak global population differentiation at both mitochondrial and nuclear loci compared with all other tropical sulids. We found evidence of high levels of gene flow between colonies within Mexico and between colonies within the southern portion of the range, but reduced gene flow between these regions. We also found evidence for population growth, isolation by distance and weak phylogeographic structure. Main conclusions Tropical seabirds can exhibit weak genetic differentiation across large geographic distances, and blue‐footed boobies exhibit the weakest population differentiation of any tropical sulid studied thus far. The weak population genetic structure that we detected in blue‐footed boobies may be caused by increased dispersal, and subsequently increased gene flow, compared with other sulids. Increased dispersal by blue‐footed boobies may be the result of the selective pressures associated with cold‐water upwelling systems, to which blue‐footed boobies appear specialized. Consideration of foraging environment may be particularly important in future studies of marine biogeography.     

Unsupervised learning and temporal context to recall complex robot trajectories

An unsupervised neural network is proposed to learn and recall complex robot trajectories. Two cases are considered: (i) A single trajectory in which a particular arm configuration (state) may occur more than once, and (ii) trajectories sharing states with each other. Ambiguities occur in both cases during recall of such trajectories. The proposed model consists of two groups of synaptic weights trained by competitive and Hebbian learning laws. They are responsible for encoding spatial and temporal features of the input sequences, respectively. Three mechanisms allow the network to deal with repeated or shared states: local and global context units, neurons disabled from learning, and redundancy. The network reproduces the current and the next state of the learned sequences and is able to resolve ambiguities. The model was simulated over various sets of robot trajectories in order to evaluate learning and recall, trajectory sampling effects and robustness.     

CoordConv-Unet: Investigating CoordConv for Organ Segmentation

ObjectivesConvolutional neural networks (CNNs) have established state-of-the-art performance in computer vision tasks such as object detection and segmentation. One of the major remaining challenges concerns their ability to capture consistent spatial and anatomically plausible attributes in medical image segmentation. To address this issue, many works advocate to integrate prior information at the level of the loss function. However, prior-based losses often suffer from local solutions and training instability. The CoordConv layers are extensions of convolutional neural network wherein convolution is conditioned on spatial coordinates. The objective of this paper is to investigate CoordConv as a proficient substitute to convolutional layers for medical image segmentation tasks when trained under prior-based losses.MethodsThis work introduces CoordConv-Unet which is a novel structure that can be used to accommodate training under anatomical prior losses. The proposed architecture demonstrates a dual role relative to prior constrained CNN learning: it either demonstrates a regularizing role that stabilizes learning while maintaining system performance, or improves system performance by allowing the learning to be more stable and to evade local minima.ResultsTo validate the performance of the proposed model, experiments are conducted on two well-known public datasets from the Decathlon challenge: a mono-modal MRI dataset dedicated to segmentation of the left atrium, and a CT image dataset whose objective is to segment the spleen, an organ characterized with varying size and mild convexity issues.ConclusionResults show that, despite the inadequacy of CoordConv when trained with the regular dice baseline loss, the proposed CoordConv-Unet structure can improve significantly model performance when trained under anatomically constrained prior losses.     

Using federated data sources and Varian Learning Portal framework to train a neural network model for automatic organ segmentation

PurposeIn this study we trained a deep neural network model for female pelvis organ segmentation using data from several sites without any personal data sharing. The goal was to assess its prediction power compared with the model trained in a centralized manner.MethodsVarian Learning Portal (VLP) is a distributed machine learning (ML) infrastructure enabling privacy-preserving research across hospitals from different regions or countries, within the framework of a trusted consortium. Such a framework is relevant in the case when there is a high level of trust among the participating sites, but there are legal restrictions which do not allow the actual data sharing between them. We trained an organ segmentation model for the female pelvic region using the synchronous data distributed framework provided by the VLP.ResultsThe prediction performance of the model trained using the federated framework offered by VLP was on the same level as the performance of the model trained in a centralized manner where all training data was pulled together in one centre.ConclusionsVLP infrastructure can be used for GPU-based training of a deep neural network for organ segmentation for the female pelvic region. This organ segmentation instance is particularly difficult due to the high variation in the organs’ shape and size. Being able to train the model using data from several clinics can help, for instance, by exposing the model to a larger range of data variations. VLP framework enables such a distributed training approach without sharing protected health information.     

Deep Learning of Orthographic Representations in Baboons

What is the origin of our ability to learn orthographic knowledge? We use deep convolutional networks to emulate the primate''s ventral visual stream and explore the recent finding that baboons can be trained to discriminate English words from nonwords [1]. The networks were exposed to the exact same sequence of stimuli and reinforcement signals as the baboons in the experiment, and learned to map real visual inputs (pixels) of letter strings onto binary word/nonword responses. We show that the networks'' highest levels of representations were indeed sensitive to letter combinations as postulated in our previous research. The model also captured the key empirical findings, such as generalization to novel words, along with some intriguing inter-individual differences. The present work shows the merits of deep learning networks that can simulate the whole processing chain all the way from the visual input to the response while allowing researchers to analyze the complex representations that emerge during the learning process.     

A neural network model for limb trajectory formation

This paper deals with the problem of representing and generating unconstrained aiming movements of a limb by means of a neural network architecture. The network produced time trajectories of a limb from a starting posture toward targets specified by sensory stimuli. Thus the network performed a sensory-motor transformation. The experimenters trained the network using a bell-shaped velocity profile on the trajectories. This type of profile is characteristic of most movements performed by biological systems. We investigated the generalization capabilities of the network as well as its internal organization. Experiments performed during learning and on the trained network showed that: (i) the task could be learned by a three-layer sequential network; (ii) the network successfully generalized in trajectory space and adjusted the velocity profiles properly; (iii) the same task could not be learned by a linear network; (iv) after learning, the internal connections became organized into inhibitory and excitatory zones and encoded the main features of the training set; (v) the model was robust to noise on the input signals; (vi) the network exhibited attractor-dynamics properties; (vii) the network was able to solve the motorequivalence problem. A key feature of this work is the fact that the neural network was coupled to a mechanical model of a limb in which muscles are represented as springs. With this representation the model solved the problem of motor redundancy.     

Taking movement data to new depths: Inferring prey availability and patch profitability from seabird foraging behavior

Detailed information acquired using tracking technology has the potential to provide accurate pictures of the types of movements and behaviors performed by animals. To date, such data have not been widely exploited to provide inferred information about the foraging habitat. We collected data using multiple sensors (GPS, time depth recorders, and accelerometers) from two species of diving seabirds, razorbills (Alca torda, N = 5, from Fair Isle, UK) and common guillemots (Uria aalge, N = 2 from Fair Isle and N = 2 from Colonsay, UK). We used a clustering algorithm to identify pursuit and catching events and the time spent pursuing and catching underwater, which we then used as indicators for inferring prey encounters throughout the water column and responses to changes in prey availability of the areas visited at two levels: individual dives and groups of dives. For each individual dive (N = 661 for guillemots, 6214 for razorbills), we modeled the number of pursuit and catching events, in relation to dive depth, duration, and type of dive performed (benthic vs. pelagic). For groups of dives (N = 58 for guillemots, 156 for razorbills), we modeled the total time spent pursuing and catching in relation to time spent underwater. Razorbills performed only pelagic dives, most likely exploiting prey available at shallow depths as indicated by the vertical distribution of pursuit and catching events. In contrast, guillemots were more flexible in their behavior, switching between benthic and pelagic dives. Capture attempt rates indicated that they were exploiting deep prey aggregations. The study highlights how novel analysis of movement data can give new insights into how animals exploit food patches, offering a unique opportunity to comprehend the behavioral ecology behind different movement patterns and understand how animals might respond to changes in prey distributions.     

DRREP: deep ridge regressed epitope predictor

Introduction

The ability to predict epitopes plays an enormous role in vaccine development in terms of our ability to zero in on where to do a more thorough in-vivo analysis of the protein in question. Though for the past decade there have been numerous advancements and improvements in epitope prediction, on average the best benchmark prediction accuracies are still only around 60%. New machine learning algorithms have arisen within the domain of deep learning, text mining, and convolutional networks. This paper presents a novel analytically trained and string kernel using deep neural network, which is tailored for continuous epitope prediction, called: Deep Ridge Regressed Epitope Predictor (DRREP).

Results

DRREP was tested on long protein sequences from the following datasets: SARS, Pellequer, HIV, AntiJen, and SEQ194. DRREP was compared to numerous state of the art epitope predictors, including the most recently published predictors called LBtope and DMNLBE. Using area under ROC curve (AUC), DRREP achieved a performance improvement over the best performing predictors on SARS (13.7%), HIV (8.9%), Pellequer (1.5%), and SEQ194 (3.1%), with its performance being matched only on the AntiJen dataset, by the LBtope predictor, where both DRREP and LBtope achieved an AUC of 0.702.

Conclusion

DRREP is an analytically trained deep neural network, thus capable of learning in a single step through regression. By combining the features of deep learning, string kernels, and convolutional networks, the system is able to perform residue-by-residue prediction of continues epitopes with higher accuracy than the current state of the art predictors.

     

The three-dimensional flight of red-footed boobies: adaptations to foraging in a tropical environment?

In seabirds a broad variety of morphologies, flight styles and feeding methods exist as an adaptation to optimal foraging in contrasted marine environments for a wide variety of prey types. Because of the low productivity of tropical waters it is expected that specific flight and foraging techniques have been selected there, but very few data are available. By using five different types of high-precision miniaturized logger (global positioning systems, accelerometers, time depth recorders, activity recorders, altimeters) we studied the way a seabird is foraging over tropical waters. Red-footed boobies are foraging in the day, never foraging at night, probably as a result of predation risks. They make extensive use of wind conditions, flying preferentially with crosswinds at median speed of 38 km h(-1), reaching highest speeds with tail winds. They spent 66% of the foraging trip in flight, using a flap-glide flight, and gliding 68% of the flight. Travelling at low costs was regularly interrupted by extremely active foraging periods where birds are very frequently touching water for landing, plunge diving or surface diving (30 landings h(-1)). Dives were shallow (maximum 2.4 m) but frequent (4.5 dives h(-1)), most being plunge dives. While chasing for very mobile prey like flying fishes, boobies have adopted a very active and specific hunting behaviour, but the use of wind allows them to reduce travelling cost by their extensive use of gliding. During the foraging and travelling phases birds climb regularly to altitudes of 20-50 m to spot prey or congeners. During the final phase of the flight, they climb to high altitudes, up to 500 m, probably to avoid attacks by frigatebirds along the coasts. This study demonstrates the use by boobies of a series of very specific flight and activity patterns that have probably been selected as adaptations to the conditions of tropical waters.     

Fully automatic and robust segmentation of the clinical target volume for radiotherapy of breast cancer using big data and deep learning

PurposeTo train and evaluate a very deep dilated residual network (DD-ResNet) for fast and consistent auto-segmentation of the clinical target volume (CTV) for breast cancer (BC) radiotherapy with big data.MethodsDD-ResNet was an end-to-end model enabling fast training and testing. We used big data comprising 800 patients who underwent breast-conserving therapy for evaluation. The CTV were validated by experienced radiation oncologists. We performed a fivefold cross-validation to test the performance of the model. The segmentation accuracy was quantified by the Dice similarity coefficient (DSC) and the Hausdorff distance (HD). The performance of the proposed model was evaluated against two different deep learning models: deep dilated convolutional neural network (DDCNN) and deep deconvolutional neural network (DDNN).ResultsMean DSC values of DD-ResNet (0.91 and 0.91) were higher than the other two networks (DDCNN: 0.85 and 0.85; DDNN: 0.88 and 0.87) for both right-sided and left-sided BC. It also has smaller mean HD values of 10.5 mm and 10.7 mm compared with DDCNN (15.1 mm and 15.6 mm) and DDNN (13.5 mm and 14.1 mm). Mean segmentation time was 4 s, 21 s and 15 s per patient with DDCNN, DDNN and DD-ResNet, respectively. The DD-ResNet was also superior with regard to results in the literature.ConclusionsThe proposed method could segment the CTV accurately with acceptable time consumption. It was invariant to the body size and shape of patients and could improve the consistency of target delineation and streamline radiotherapy workflows.     

DeepMIB: User-friendly and open-source software for training of deep learning network for biological image segmentation

We present DeepMIB, a new software package that is capable of training convolutional neural networks for segmentation of multidimensional microscopy datasets on any workstation. We demonstrate its successful application for segmentation of 2D and 3D electron and multicolor light microscopy datasets with isotropic and anisotropic voxels. We distribute DeepMIB as both an open-source multi-platform Matlab code and as compiled standalone application for Windows, MacOS and Linux. It comes in a single package that is simple to install and use as it does not require knowledge of programming. DeepMIB is suitable for everyone interested of bringing a power of deep learning into own image segmentation workflows.     

Dynamic graph convolution neural network based on spatial-temporal correlation for air quality prediction

Air pollution is a serious threat to both the ecological environment and the physical health of individuals. Therefore, accurate air quality prediction is urgent and necessary for pollution mitigation and residents’ travel. However, few existing models are established based on the dynamic spatiotemporal correlation of air pollutants to predict air quality. In this paper, a novel deep learning model combining the dynamic graph convolutional network and the multi-channel temporal convolutional network (DGC-MTCN) is proposed for air quality prediction. To efficiently represent the time-varying spatial dependencies, a new spatiotemporal dynamic correlation calculation method based on gray relation analysis is proposed to construct dynamic adjacency matrices. Then, the spatiotemporal features are sufficiently extracted by the graph convolutional network and the multi-channel temporal convolutional network. Two real-world air quality datasets collected from Beijing and Fushun are applied to verify the performance of our proposed model. The experimental results show that compared with other baselines, the DGC-MTCN model has excellent prediction accuracy. Especially for the prediction of multi-step and different stations, our model performs better temporal stability and generalization ability.     

Energy-based graph convolutional networks for scoring protein docking models

Structural information about protein-protein interactions, often missing at the interactome scale, is important for mechanistic understanding of cells and rational discovery of therapeutics. Protein docking provides a computational alternative for such information. However, ranking near-native docked models high among a large number of candidates, often known as the scoring problem, remains a critical challenge. Moreover, estimating model quality, also known as the quality assessment problem, is rarely addressed in protein docking. In this study, the two challenging problems in protein docking are regarded as relative and absolute scoring, respectively, and addressed in one physics-inspired deep learning framework. We represent protein and complex structures as intra- and inter-molecular residue contact graphs with atom-resolution node and edge features. And we propose a novel graph convolutional kernel that aggregates interacting nodes’ features through edges so that generalized interaction energies can be learned directly from 3D data. The resulting energy-based graph convolutional networks (EGCN) with multihead attention are trained to predict intra- and inter-molecular energies, binding affinities, and quality measures (interface RMSD) for encounter complexes. Compared to a state-of-the-art scoring function for model ranking, EGCN significantly improves ranking for a critical assessment of predicted interactions (CAPRI) test set involving homology docking; and is comparable or slightly better for Score_set, a CAPRI benchmark set generated by diverse community-wide docking protocols not known to training data. For Score_set quality assessment, EGCN shows about 27% improvement to our previous efforts. Directly learning from 3D structure data in graph representation, EGCN represents the first successful development of graph convolutional networks for protein docking.     

Summer diving and haul‐out behavior of leopard seals (Hydrurga leptonyx) near mesopredator breeding colonies at Livingston Island,Antarctic Peninsula

Leopard seals are conspicuous apex predators in Antarctic coastal ecosystems, yet their foraging ecology is poorly understood. Historically, the ecology of diving vertebrates has been studied using high‐resolution time‐depth records; however, to date such data have not been available for leopard seals. Twenty‐one time‐depth recorders were deployed on seasonally resident adult females in January and February between 2008 and 2014. The average deployment length was 13.65 ± 11.45 d and 40,308 postfilter dives were recorded on 229 foraging trips. Dive durations averaged 2.20 ± 1.23 min. Dives were shallow with 90.1% measuring 30 m or less, and a mean maximum dive depth of 16.60 ± 10.99 m. Four dive types were classified using a k‐means cluster analysis and compared with corresponding animal‐borne video data. Dive activity (number of dives/hour) was concentrated at night, including crepuscular periods. Haul‐out probabilities were highest near midday and were positively correlated with available daylight. Visual observations and comparisons of diving activity between and within years suggest individual‐based differences of foraging effort by time of day. Finally, dive and video data indicate that in addition to at‐surface hunting, benthic searching and facultative scavenging are important foraging strategies for leopard seals near coastal mesopredator breeding colonies.     

Automatic strawberry leaf scorch severity estimation via faster R-CNN and few-shot learning

Using deep learning to estimate strawberry leaf scorch severity often achieves unsatisfactory results when a strawberry leaf image contains complex background information or multi-class diseased leaves and the number of annotated strawberry leaf images is limited. To solve these issues, in this paper, we propose a two-stage method including object detection and few-shot learning to estimate strawberry leaf scorch severity. In the first stage, Faster R-CNN is used to mark the location of strawberry leaf patches, where each single strawberry leaf patch is clipped from original strawberry leaf images to compose a new strawberry leaf patch dataset. In the second stage, the Siamese network trained on the new strawberry leaf patch dataset is used to identify the strawberry leaf patches and then estimate the severity of the original strawberry leaf scorch images according to the multi-instance learning concept. Experimental results from the first stage show that Faster R-CNN achieves better mAP in strawberry leaf patch detection than other object detection networks, at 94.56%. Results from the second stage reveal that the Siamese network achieves an accuracy of 96.67% in the identification of strawberry disease leaf patches, which is higher than the Prototype network. Comprehensive experimental results indicate that compared with other state-of-the-art models, our proposed two-stage method comprising the Faster R-CNN (VGG16) and Siamese networks achieves the highest estimation accuracy of 96.67%. Moreover, our trained two-stage model achieves an estimation accuracy of 88.83% on a new dataset containing 60 strawberry leaf images taken in the field, which indicates its excellent generalization ability.     

Comparative phylogeography of brown (Sula leucogaster) and red-footed boobies (S. sula): The influence of physical barriers and habitat preference on gene flow in pelagic seabirds

To test the hypothesis that both physical and ecological barriers to gene flow drive population differentiation in tropical seabirds, we surveyed mitochondrial control region variation in 242 brown boobies (Sula leucogaster), which prefer inshore habitat, and 271 red-footed boobies (S. sula), which prefer pelagic habitat. To determine the relative influence of isolation and gene flow on population structure, we used both traditional methods and a recently developed statistical method based on coalescent theory and Bayesian inference (Isolation with Migration). We found that global population genetic structure was high in both species, and that female-mediated gene flow among ocean basins apparently has been restricted by major physical barriers including the Isthmus of Panama, and the periodic emergence of the Sunda and Sahul Shelves in Southeast Asia. In contrast, the evolutionary history of populations within ocean basins differed markedly between the two species. In brown boobies, we found high levels of population genetic differentiation and limited gene flow among colonies, even at spatial scales as small as 500 km. Although red-footed booby colonies were also genetically differentiated within ocean basins, coalescent analyses indicated that populations have either diverged in the face of ongoing gene flow, or diverged without gene flow but recently made secondary contact. Regardless, gene flow among red-footed booby populations was higher than among brown booby populations. We suggest that these contrasting patterns of gene flow within ocean basins may be explained by the different habitat preferences of brown and red-footed boobies.     

Neural hierarchical models of ecological populations

Neural networks are increasingly being used in science to infer hidden dynamics of natural systems from noisy observations, a task typically handled by hierarchical models in ecology. This article describes a class of hierarchical models parameterised by neural networks – neural hierarchical models. The derivation of such models analogises the relationship between regression and neural networks. A case study is developed for a neural dynamic occupancy model of North American bird populations, trained on millions of detection/non‐detection time series for hundreds of species, providing insights into colonisation and extinction at a continental scale. Flexible models are increasingly needed that scale to large data and represent ecological processes. Neural hierarchical models satisfy this need, providing a bridge between deep learning and ecological modelling that combines the function representation power of neural networks with the inferential capacity of hierarchical models.     

Improving cerebral microvascular image quality of optical coherence tomography angiography with deep learning-based segmentation

Optical coherence tomography angiography (OCTA) can map the microvascular networks of the cerebral cortices with micrometer resolution and millimeter penetration. However, the high scattering of the skull and the strong noise in the deep imaging region will distort the vasculature projections and decrease the OCTA image quality. Here, we proposed a deep learning-based segmentation method based on a U-Net convolutional neural network to extract the cortical region from the OCT image. The vascular networks were then visualized by three OCTA algorithms. The image quality of the vasculature projections was assessed by two metrics, including the peak signal-to-noise ratio (PSNR) and the contrast-to-noise ratio (CNR). The results show the accuracy of the cortical segmentation was 96.07%. The PSNR and CNR values increased significantly in the projections of the selected cortical regions. The OCTA incorporating the deep learning-based cortical segmentation can efficiently improve the image quality and enhance the vasculature clarity.     

Tumor Cell Identification in Ki-67 Images on Deep Learning

The proportion of cells staining for the nuclear antigen Ki-67 is an important predictive indicator for assessment of tumor cell proliferation and growth in routine pathological investigation. Instead of traditional scoring methods based on the experience of a trained laboratory scientist, deep learning approach can be automatically used to analyze the expression of Ki-67 as well. Deep learning based on convolutional neural networks (CNN) for image classification and single shot multibox detector (SSD) for object detection are used to investigate the expression of Ki-67 for assessment of biopsies from patients with breast cancer in this study. The results focus on estimating the probability heatmap of tumor cells using CNN with accuracy of 98% and detecting the tumor cells using SSD with accuracy of 90%. This deep learning framework will provide an objective basis for the malignant degree of breast tumors and be beneficial to the pathologists for fast and efficiently Ki-67 scoring.