DW-Net: Dynamic Multi-Hierarchical Weighting Segmentation Network for Joint Segmentation of Retina Layers With Choroid Neovascularization

Choroid neovascularization (CNV) is one of the blinding factors. The early detection and quantitative measurement of CNV are crucial for the establishment of subsequent treatment. Recently, many deep learning-based methods have been proposed for CNV segmentation. However, CNV is difficult to be segmented due to the complex structure of the surrounding retina. In this paper, we propose a novel dynamic multi-hierarchical weighting segmentation network (DW-Net) for the simultaneous segmentation of retinal layers and CNV. Specifically, the proposed network is composed of a residual aggregation encoder path for the selection of informative feature, a multi-hierarchical weighting connection for the fusion of detailed information and abstract information, and a dynamic decoder path. Comprehensive experimental results show that our proposed DW-Net achieves better performance than other state-of-the-art methods.

Computational mechanisms of distributed value representations and mixed learning strategies

Learning appropriate representations of the reward environment is challenging in the real world where there are many options, each with multiple attributes or features. Despite existence of alternative solutions for this challenge, neural mechanisms underlying emergence and adoption of value representations and learning strategies remain unknown. To address this, we measure learning and choice during a multi-dimensional probabilistic learning task in humans and trained recurrent neural networks (RNNs) to capture our experimental observations. We find that human participants estimate stimulus-outcome associations by learning and combining estimates of reward probabilities associated with the informative feature followed by those of informative conjunctions. Through analyzing representations, connectivity, and lesioning of the RNNs, we demonstrate this mixed learning strategy relies on a distributed neural code and opponency between excitatory and inhibitory neurons through value-dependent disinhibition. Together, our results suggest computational and neural mechanisms underlying emergence of complex learning strategies in naturalistic settings.

Informative sign of determining the category of transmitted data in high-speed information transmission networks.

Difficulties arise with the recognition of the traffic category in the conditions of widespread use in telecommunication networks of many different non-standardized data transmission protocols. The article discusses the features of data transmission for the category of real-time data and data not critical to delays. As an informative feature for recognizing a data category, it is proposed to use an estimate of the bit rate of a single subscriber. The paper presents the results of an experiment to estimate the bit rate of a single subscriber of the assessment, substantiate the distribution laws of estimates, and determine the threshold value of a feature for recognizing a traffic category.

SE-Mask R-CNN: An improved Mask R-CNN for apple detection and segmentation

Fruit detection and segmentation is an essential operation of orchard yield estimation, the result of yield estimation directly depends on the speed and accuracy of detection and segmentation. In this work, we propose an effective method based on Mask R-CNN to detect and segment apples under complex environment of orchard. Firstly, the squeeze-and-excitation block is introduced into the ResNet-50 backbone, which can distribute the available computational resources to the most informative feature map in channel-wise. Secondly, the aspect ratio is introduced into the bounding box regression loss, which can promote the regression of bounding boxes by deforming the shape of bounding boxes to the apple boxes. Finally, we replace the NMS operation in Mask R-CNN by Soft-NMS, which can remove the redundant bounding boxes and obtain the correct detection results reasonably. The experimental result on the Minneapple dataset demonstrates that our method overperform several state-of-the-art on apple detection and segmentation.

FRACTAL ANALYSIS AS A TOOL FOR AUTOMATIC DETECTION OF ARTIFACTS IN ELECTROENCEPHALOGRAM RECORDS

В статье обсуждается применение методов фрактального анализа для решения задачи автоматической фильтрации сигнала ЭЭГ от артефактов различной природы. Изучается возможность использования показателя Херста в качестве информативного признака для алгоритмов интеллектуальной обработки данных. The article discusses the possibility of using fractal analysis to solve the problem of automatic filtering of the EEG signal from artifacts of various nature. The possibility of using the Hurst exponent as an informative feature for intelligent data processing algorithms is investigated

An Improved RF-PNN Based Annotation Model With Texture and Edge Based Feature Descriptors

High informative feature descriptors always improves the classification process. In order to classify the earth surface, it is essential to annotate satellite images using itshigh informative feature descriptors. In this proposed work, an annotation framework has been implemented to improve the image discrimination by extracting texture and edge based feature vectors. So the combination of these features subsequently fed into the Random Forest based Probability Neural Network (RF-PNN) classifier to make an annotation model. The experimental analysis with comparisons shows that the proposed annotation model well performed with earlier works and comparative results of benchmark datasets of AID dataset, UC-Merced Land-Use dataset and WHU-RS19 datasets have been documented with analysis.

Time-resolved structure-function coupling in brain networks

The relationship between structural and functional connectivity in the brain is a key question in systems neuroscience. Modern accounts assume a single global structure-function relationship that persists over time. Here we show that structure-function coupling is dynamic and regionally heterogeneous. We use a temporal unwrapping procedure to identify moment-to-moment co-fluctuations in neural activity, and reconstruct time-resolved structure-function coupling patterns. We find that patterns of dynamic structure-function coupling are highly organized across the cortex. These patterns reflect cortical hierarchies, with stable coupling in unimodal and transmodal cortex, and dynamic coupling in intermediate regions, particularly in insular cortex (salience network) and frontal eye fields (dorsal attention network). Finally, we show that the variability of structure-function coupling is shaped by the distribution of connection lengths. The time-varying coupling of structural and functional connectivity points towards an informative feature of the brain that may reflect how cognitive functions are flexibly deployed and implemented.