Coarse Network Research Articles

Mobile Observation of Air Temperature and Humidity Distributions under Summer Sea Breezes in the Central Area of Osaka City

Thermal environment of urban areas in the summertime has become harmful to human health due to global warming and the urban heat island (UHI) effect. Mobile observations enable us to obtain the distribution of air temperature at microscale, such as urban blocks, which cannot be captured by the coarse network of meteorological sites. A series of mobile measurements was executed in the central area of Osaka City in Japan, around the Nakanoshima district which lies between two rivers, to investigate the air temperature and humidity distributions in a built-up area under sea breeze conditions. Upper wind and surface temperature of the rivers were also observed using pilot balloons and infrared thermography camera, respectively. The mean air temperature in Nakanoshima was generally lower than that of the surrounding area. Urban geometries such as building density and building height seem to affect the mean air temperature by changing the ventilation efficiency. Humidity was inversely correlated with air temperature distribution but was higher at the confluence of rivers and green parks. The depth of the sea breeze layer was found to be about 1 km. Sea breezes close to the ground surface penetrated the city along the rivers, sandwiching the Nakanoshima district. During the daytime, the surface temperatures of the rivers were lower than the air temperature observed at the nearest stationary observation point, and the difference reached approximately 2 °C.

Image Inpainting Using Two-Stage Loss Function and Global and Local Markovian Discriminators.

Image inpainting networks can produce visually reasonable results in the damaged regions. However, existing inpainting networks may fail to reconstruct the proper structures or tend to generate the results with color discrepancy. To solve this issue, this paper proposes an image inpainting approach using the proposed two-stage loss function. The loss function consists of different Gaussian kernels, which are utilized in different stages of network. The use of our two-stage loss function in coarse network helps to focus on the image structure, while the use of it in refinement network is helpful to restore the image details. Moreover, we proposed a global and local PatchGANs (GAN means generative adversarial network), named GL-PatchGANs, in which the global and local markovian discriminators were used to control the final results. This is beneficial to focus on the regions of interest (ROI) on different scales and tends to produce more realistic structural and textural details. We trained our network on three popular datasets on image inpainting separately, both Peak Signal to Noise ratio (PSNR) and Structural Similarity (SSIM) between our results, and ground truths on test images show that our network can achieve better performance compared with the recent works in most cases. Besides, the visual results on three datasets also show that our network can produce visual plausible results compared with the recent works.

Network theory of the bacterial ribosome.

In the past two decades, research into the biochemical, biophysical and structural properties of the ribosome have revealed many different steps of protein translation. Nevertheless, a complete understanding of how they lead to a rapid and accurate protein synthesis still remains a challenge. Here we consider a coarse network analysis in the bacterial ribosome formed by the connectivity between ribosomal (r) proteins and RNAs at different stages in the elongation cycle. The ribosomal networks are found to be dis-assortative and small world, implying that the structure allows for an efficient exchange of information between distant locations. An analysis of centrality shows that the second and fifth domains of 23S rRNA are the most important elements in all of the networks. Ribosomal protein hubs connect to much fewer nodes but are shown to provide important connectivity within the network (high closeness centrality). A modularity analysis reveals some of the different functional communities, indicating some known and some new possible communication pathways Our mathematical results confirm important communication pathways that have been discussed in previous research, thus verifying the use of this technique for representing the ribosome, and also reveal new insights into the collective function of ribosomal elements.

AUTOPHAGIC MYOPATHIES / MYOFIBRILLAR MYOPATHIES / DISTAL MYOPATHIES / POMPE DISEASE: P.12 Mutations in the supervillin gene cause myopathy with myofibrillar disorganization and autophagic vacuoles

We have identified four patients with a childhood/adolescence onset of a myopathy from two unrelated consanguineous families. The patients have a wide neck, anteverted shoulders and prominent trapezius muscles together with variable contractures as characteristic features. All patients showed increased levels of serum creatine kinase but no or minor muscle weakness. Mild cardiac manifestations were observed. Light and electron microscopical investigations revealed a structural myopathy with numerous lobulated muscle fibers and considerable myofibrillar alterations with a coarse and irregular intermyofibrillar network.

Observational Practices for Urban Microclimates Using Meteorologically Instrumented Unmanned Aircraft Systems

The urban boundary layer (UBL) is one of the most important and least understood atmospheric domains and, consequently, warrants deep understanding and rigorous analysis via sophisticated experimental and numerical tools. When field experiments have been undertaken, they have primarily been accomplished with either a coarse network of in-situ sensors or slow response sensors based on timing or Doppler shifts, resulting in low resolution and decreasing performance with height. Small unmanned aircraft systems (UASs) offer an opportunity to improve on traditional UBL observational strategies that may require substantive infrastructure or prove impractical in a vibrant city, prohibitively expensive, or coarse in resolution. Multirotor UASs are compact, have the ability to take-off and land vertically, hover for long periods of time, and maneuver easily in all three spatial dimensions, making them advantageous for probing an obstacle-laden environment. Fixed-wing UASs offer an opportunity to cover vast horizontal and vertical distances, at low altitudes, in a continuous manner with high spatial resolution. Hence, fixed-wing UASs are advantageous for observing the roughness sublayer above the highest building height where traditional manned aircraft cannot safely fly. This work presents a methodology for UBL investigations using meteorologically instrumented UASs and discusses lessons learned and best practices garnered from a proof of concept field campaign that focused on the urban canopy layer and roughness sublayer of a large modern city with a high-rise urban canopy.

Coarse-to-fine classification for diabetic retinopathy grading using convolutional neural network.

Diabetic retinopathy (DR) is the most common eye complication of diabetes and one of the leading causes of blindness and vision impairment. Automated and accurate DR grading is of great significance for the timely and effective treatment of fundus diseases. Current clinical methods remain subject to potential time-consumption and high-risk. In this paper, a hierarchically Coarse-to-fine network (CF-DRNet) is proposed as an automatic clinical tool to classify five stages of DR severity grades using convolutional neural networks (CNNs). The CF-DRNet conforms to the hierarchical characteristic of DR grading and effectively improves the classification performance of five-class DR grading, which consists of the following: (1) The Coarse Network performs two-class classification including No DR and DR, where the attention gate module highlights the salient lesion features and suppresses irrelevant background information. (2) The Fine Network is proposed to classify four stages of DR severity grades of the grade DR from the Coarse Network including mild, moderate, severe non-proliferative DR (NPDR) and proliferative DR (PDR). Experimental results show that proposed CF-DRNet outperforms some state-of-art methods in the publicly available IDRiD and Kaggle fundus image datasets. These results indicate our method enables an efficient and reliable DR grading diagnosis in clinic.

Tuning the microstructure morphology and genetic mechanical properties of 2219 Al alloy with ultrasonic treatment

Some defects i.e. dendrite, coarse eutectic phases, and serious macro-segregation were hard to eliminate when the ultra-large aluminum alloy was manufactured by conventional technique. An effective method with ultrasonic sonotrode was inserted into the direct-cooling casting system to manufacture the 2219 Al alloy billet (Φ1100 mm × 5400 mm), followed by multidirectional forging and solution-aging (MSA) treatment. A comparative study of the morphology of α-Al grains, eutectic phases, and precipitates, as well as the mechanical properties were investigated. The results indicated that more coarse eutectic networks agglomerated together in the center position compared with billet samples without ultrasonic treatment (NUT) and with ultrasonic treatment (UT). They were difficult to completely dissolve into the matrix in the downstream MSA process, and acted as preferential nucleation sites for micro-cracks, thereby decreasing the mechanical properties. The average grain sizes of billet were generally refined induced by UT. And a greater density of tinier, uniform distributed precipitates was obtained from the UT billet and MSA samples. The mechanical properties were obviously increased by ultrasonic treatment in the billet samples. Based on the present results and proposed hypotheses in previous works, the dominant mechanisms of UT on major microstructures and genetic effects on mechanical properties were discussed.

Generation of High-resolution Lung Computed Tomography Images using Generative Adversarial Networks.

To deal with the limiting data in training for new deep learning modules, we purpose a method to generate high-resolution medical images by implementing generative adversarial networks (GAN) models. Firstly, the boundary equilibrium generative adversarial networks model was used to generate the whole lung computed tomography images. Image inpainting was then integrated to generate the delicate details of the lung part by dividing into a coarse network and a refinement network to inpaint more completed and intricate details. With this method, we aim to increase the amount of high-resolution medical images for future applications in deep learning.

Change of network structure in agarose gels by aging during storage studied by NMR and electrophoresis

Changes of the network structure in agarose gels during storage were studied by measuring diffusion coefficient (D) of probe polymers. PGSTE 1H NMR was used to measure the D of pullulan added as a probe to show an increase with storage suggesting the formation of coarse network with thicker bundles of agarose aggregates. The hydrodynamic mesh size was estimated from the degree of restriction in the diffusion shedding light on the change of microscopic environment in the agarose gel during storage. From the gel electrophoresis, D of DNA was calculated from the friction on the DNA movement under the electric field to show an increase of D with storage suggesting the change of network structure in the aged agarose gels. The results provided a better understanding of the aging mechanism in agarose gels with the conformational changes which should be commonly observed in polysaccharides.

Effect of protein aggregation on rheological properties of pea protein gels

Yoghurt style gels are a promising way to increase the consumption of plant derived proteins. However, reaching texture properties similar to those commonly known from milk yoghurts while incorporating large amounts of plant-derived proteins, is a challenge that needs to be addressed to meet consumers expectations. Therefore, this study aims to investigate the influence of pH conditions (pH 6.0 to pH 8.0) during pre-treatment on the rheological properties of fermentation induced pea protein gels with a protein content of 10%. Results showed a strong correlation between the pH value during pre-treatment and the protein solubility after pH readjustment to pH 8. Solubility was highest if pea protein was pre-treated at pH 8.0 and lowest if it was pre-treated at pH 6.0. Since only soluble aggregates are believed to participate in network formation, networks formed by pea protein pre-treated at pH 6 were coarser, than those formed by pea protein pre-treated at pH 8 owing to a lower degree of crosslinking caused by less available protein. Coarser networks and higher proportions of insoluble particles increased loss of water and lowered the storage modulus G’ as well as the ability of the networks to recover after intense shearing. In particular pre-treatment at pH values below 7.0 led to gels with storage moduli of the same magnitude as those measured in commercial milk derived yoghurts. Adjusting the pH value during pre-treatment of pea protein can therefore be considered a promising approach for the customisation of texture properties while maintaining a constantly high protein content.

Spatiotemporal Heterogeneity of κ-Carrageenan Gels Investigated via Single-Particle-Tracking Fluorescence Microscopy.

Hydrogels made of the polysaccharide κ-carrageenan are widely used in the food and personal care industry as thickeners or gelling agents. These hydrogels feature dense regions embedded in a coarser bulk network, but the characteristic size and behavior of these regions have remained elusive. Here, we use single-particle-tracking fluorescence microscopy (sptFM) to quantitatively describe κ-carrageenan gels. Infusing fluorescent probes into fully gelated κ-carrageenan hydrogels resulted in two distinct diffusional behaviors. Obstructed self-diffusion of the probes revealed that the coarse network consists of κ-carrageenan strands with a typical diameter of 3.2 ± 0.3 nm leading to a nanoprobe diffusion coefficient of ∼1–5 × 10–12 m2/s. In the dense network regions, we found a fraction with a largely decreased diffusion coefficient of ∼1 × 10–13 m2/s. We also observed dynamic exchange between these states. The computation of spatial mobility maps from the diffusional data indicated that the dense network regions have a characteristic diameter of ∼1 μm and show mobility on the second-to-minute timescale. sptFM provides an unprecedented view of spatiotemporal heterogeneity of hydrogel networks, which we believe bears general relevance for understanding transport and release of both low- and high-molecular weight solutes.

LookCom: Learning Optimal Network for Community Detection

Community detection is one of the fundamental tasks in graph mining, which aims to identify group assignment of nodes in a complex network. Recently, network embedding techniques have demonstrated their strong power in advancing the community detection task and achieve better performance than various traditional methods. Despite their empirical success, most of the existing algorithms directly leverage the observed coarse network structure for community detection. Therefore, they often lead to suboptimal performance as the observed connections fail to capture the essential tie strength information among nodes precisely and account for the impact of noisy links. In this paper, an optimal network structure for community detection is introduced to characterize the fine-grained tie strength information between connected nodes and alleviate the adverse effects of noisy links. To obtain an expressive node representation for community detection, we learn the optimal network structure and network embeddings in a joint framework, instead of using a two-stage approach to derive the node embeddings from the coarse network topology. In particular, we formulate the joint framework as an optimization problem and an alternating optimization algorithm is exploited to solve the proposed optimization problem. Additionally, theoretical analyses regarding the computational complexity and the convergence of the optimization algorithm are also provided. Extensive experiments on both synthetic and real-world networks demonstrate the effectiveness and superiority of the proposed framework.

Accelerated Carbon Atoms Diffusion in Bearing Steel Using Electropulsing to Reduce Spheroidization Processing Time and Improve Microstructure Uniformity

Herein, the thermodynamic and kinetic limitations of carbide dissolution behavior during the traditional heat treatment process are broken through the external field treatment. The difference in electrical conductivity between the matrix and the carbide affects the electric current distribution, which ultimately results in electric field distortion. The numerical calculation results show that under the coupled field of heat and pulsed electric current, the distribution of the current density around the particles is related to the size of the particles. The selective distribution of current caused the energy difference to eventually promote the preferential dissolution of coarse network carbides at grain boundaries. The redistribution of carbon atoms between the grain boundaries and the grain interior provides a good initial structure for the subsequent precipitation process, and finally, uniform precipitation of cementite is achieved in the matrix. Meanwhile, the electromigration of carbon atoms under pulsed electric current increases the diffusion rate of carbon atom and shortens the processing time (traditional heat treatment is about 20 h, and the electropulse‐based method in this study is about 4 h).

Pose-Guided Multi-Granularity Attention Network for Text-Based Person Search

Text-based person search aims to retrieve the corresponding person images in an image database by virtue of a describing sentence about the person, which poses great potential for various applications such as video surveillance. Extracting visual contents corresponding to the human description is the key to this cross-modal matching problem. Moreover, correlated images and descriptions involve different granularities of semantic relevance, which is usually ignored in previous methods. To exploit the multilevel corresponding visual contents, we propose a pose-guided multi-granularity attention network (PMA). Firstly, we propose a coarse alignment network (CA) to select the related image regions to the global description by a similarity-based attention. To further capture the phrase-related visual body part, a fine-grained alignment network (FA) is proposed, which employs pose information to learn latent semantic alignment between visual body part and textual noun phrase. To verify the effectiveness of our model, we perform extensive experiments on the CUHK Person Description Dataset (CUHK-PEDES) which is currently the only available dataset for text-based person search. Experimental results show that our approach outperforms the state-of-the-art methods by 15 % in terms of the top-1 metric.

Strengthening and ductilization mechanisms of friction stir processed cast Mg–Al–Zn alloy

AZ91 alloy, the most widely used Mg casting alloy, exhibits low strength/ductility and weak energy absorption, which is a function of its large grain size and the presence of a coarse and continuous network of β-Mg17Al12 intermetallic compounds. This work demonstrated that friction stir processing (FSP) enables enhancement of strength and energy absorption capability of AZ91 alloy. The influence of FSP treatment on various potential strengthening mechanisms, including grain boundary, solid solution, and sub-micron particle strengthening mechanisms, was studied. It is identified that the grain boundary strengthening plays a significant contribution to the strength of the FSP treated AZ91. FSP treatment altered the failure mechanism of the alloy from brittle cleavage-dominant mode to ductile dimple-dominant mode, which can increase the potential of cast Mg alloys to use in the safety-critical application. The significant improvement in energy absorption capability is a function of grain refining and the formation of ultra-fine sub-micron Mg17Al12 particles. The trends in mechanical properties of AZ91 treated by various severe plastic deformation approaches are disused.

Filamentous and step-like behavior of gelling coarse fibrin networks revealed by high-frequency microrheology.

By a micro-experimental methodology, we study the ongoing molecular process inside coarse fibrin networks by means of microrheology. We made these networks gelate around a probe microbead, allowing us to observe a temporal evolution compatible with the well-known molecular formation of fibrin networks in four steps: monomer, protofibril, fiber and network. Thanks to the access that optical-trapping interferometry provides to the short-time scale on the bead's Brownian motion, we observe a Kelvin-Voigt mechanical behavior from low to high frequencies, range not available in conventional rheometry. We exploit that mechanical model for obtaining the characteristic lengths of the filamentous structures composing these fibrin networks, whose obtained values are compatible with a non-affine behavior characterized by bending modes. At very long gelation times, a ω7/8 power-law is observed in the loss modulus, theoretically related with the longitudinal response of the molecular structures.

S2DNet: Depth Estimation From Single Image and Sparse Samples

Depth prediction from single image is a challenging task due to the intra scale ambiguity and unavailability of prior information. The prediction of an unambiguous depth from single RGB image is very important aspect for computer vision applications. In this paper, an end-to-end sparse-to-dense network (S2DNet) is proposed for single image depth estimation (SIDE). The proposed network processes single image along with the additional sparse depth samples for depth estimation. The additional sparse depth sample are acquired either with a low-resolution depth sensor or calculated by visual simultaneous localization and mapping (SLAM) algorithms. In first stage, the proposed S2DNet estimates coarse-level depth map using sparse-to-dense coarse network (S2DCNet). In second stage, the estimated coarse-level depth map is concatenated with the input image and used as an input to the sparse-to-dense fine network (S2DFNet) for fine-level depth map estimation. The proposed S2DFNet comprises of attention map architecture which helps to estimate the prominent depth information. The quantitative and qualitative performance evaluation of the proposed network has been carried out using the error metrics. We perform complete evaluation of S2DNet on four publically available benchmark data sets i.e. NYU Depth-V2 indoor dataset [1] , KITTI odometry outdoor dataset [2] , KITTI depth completion test database [3] and SUN-RGB database [4] . Further, we have extended the proposed S2DNet for image de-hazing. The experimental analysis shows that the proposed S2DNet outperforms the existing state-of-the-art methods for both single image depth estimation and image de-hazing.

Coarse and Fine Granularity Graph Reasoning for Interpretable Multi-Hop Question Answering

Interpretable multi-hop question answering requires step-by-step reasoning over multiple documents and finding scattered supporting facts to answer the question. Prior works have proposed the entity graph method to aggregate the entity information to improve the ability of reasoning. However, the entity graph loses some non-entity information that is also important to understand the semantics. Moreover, the entities distributed in the noisy sentences may mislead the reasoning process. In this paper, we propose the Coarse and Fine Granularity Graph Network (CFGGN), a novel interpretable model that combines both sentence information and entity information to answer the multi-hop questions. The CFGGN consists of a coarse-grain module to perform sentence-level reasoning and a fine-grain module to make entity-level reference. In sentence-level reasoning, the sentence graph is constructed to filter out the noisy sentences and capture the sentence features. In entity-level reference, a dynamic entity graph is used for the entity-level reasoning. We design a fusion module to integrate information of different granularity. To enhance the interpretability of the overall process, we calculate the reasoning score for each step and present the reasoning path from the multiple documents to the final answer. Evaluation on the HotpotQA dataset in the distractor setting shows that our method outperforms the published SOTA entity-based method in five out of six metrics.

Climate monitoring and heat and cold waves detection over France using a new spatialization of daily temperature extremes from 1947 to present

Abstract. For many years real-time climate monitoring for temperature over France has been performed using a national index built by averaging the daily mean temperatures of constant subset of 30 stations with long-term series. In order to derive climate indices at finer scales, a spatialization of extreme daily temperatures (called ANASTASIA) had been produced on a 1 km regular grid using a regression-kriging method. The production covers 1947 to present period. Cross-validation shows low biases after the 1960s. The temporal homogeneity of the product is satisfying at the national scale from the 1970s. However, a high impact of the network density has been found and the use of a too coarse observation network deteriorates the analysis creating temporal heterogeneities. Finally, the ANASTASIA analysis has been used for real-time monitoring over France and detection of heat and cold wave episodes. The new products based on ANASTASIA are consistent with the current operational ones at national scale while bringing added values at local scales.

Attend to count: Crowd counting with adaptive capacity multi-scale CNNs

Crowd counting is a challenging task due to the large variations in crowd distributions. Previous methods tend to tackle the whole image with a single fixed structure, which is unable to handle diverse complicated scenes with different crowd densities. Hence, we propose the Adaptive Capacity Multi-scale convolutional neural networks (ACM-CNN), a novel crowd counting approach which can assign different capacities to different portions of the input. The intuition is that the model should focus on important regions of the input image and optimize its capacity allocation conditioning on the crowd intensive degree. ACM-CNN consists of three types of modules: A coarse network, a fine network, and a smooth network. The coarse network is used to explore the areas that need to be focused via count attention mechanism, and generate a rough feature map. Then the fine network processes the areas of interest into a fine feature map. To alleviate the sense of division caused by fusion, the smooth network is designed to combine two feature maps organically to produce high-quality density maps. Extensive experiments are conducted on five mainstream datasets. The results demonstrate the effectiveness of the proposed model for both density estimation and crowd counting tasks.