Search

The triplet loss is widely used in learning the local descriptors for image matching. However, existing triplet loss-based methods, like HardNet and DSM, employ the point-to-point distance metric, which neglects the neighborhood information of descriptors. Considering the fact that local neighborhood structures of matching descriptors would be similar under the ideal condition, this paper aims to learn the neighborhood topology-consistent descriptors (TCDesc). To this end, we first propose the linear combination weight as the topology weight to depict the neighborhood topology for each descriptor, where the difference between the center descriptor and the linear combination of its neighbors is minimized. For the global comparison, we then define a global topology vector by using the local topology weights. Next, beyond the Euclidean distance, we define a topology distance with the topology vectors to indicate the topological difference between the matching descriptors. Furthermore, we propose an adaptive weighting strategy to jointly minimize the topology distance and Euclidean distance in triplet loss. Experimental results on four widely-used datasets, i.e., UBC PhotoTourism, HPatches, W1BS and Oxford, demonstrate that our method can effectively improve the performance of both HardNet and DSM.

Dynamic threshold neural P systems (DTNP systems) are a theoretical computing model proposed in previous work. As a variant of spiking neural P (SNP) systems, DTNP systems have two mechanisms: spiking and dynamic threshold mechanisms. By considering local connections of neurons, we design two-dimensional DTNP systems with local topology. Based on the DTNP systems, we develop a novel fusion method based on Laplacian pyramid for medical images. In the decomposition layers of Laplacian pyramid, WLE and INSML features are combined as the input of DTNP systems, and its output is used as the control signal of fusion rules. The proposed fusion method is evaluated on seven pairs of benchmark medical images and is compared with eight baseline fusion methods. Experimental results demonstrate the advantage of the proposed fusion method for the fusion of medical images.

It is a highly productive approach to analyze human languages with complex networks, because the language network can effectively predict unknown features. One of the classical features in complex networks is module (or community) structure, which features dense intra-group connections and sparse inter-group connections. This work aims to identify oracle characters, the oldest characters in China, with similar semantic through module structure detection because about two-thirds of all oracle characters remain unknown. Firstly, two oracle character networks are constructed based on the context and shape of oracle rubbings and characters. Then, three module structure detection methods are employed to mine the modules from the oracle character networks. Through the experiments on the oracle character networks, three significant results are obtained. The first one is that oracle characters (or oracle variant characters) with similar semantics can be discovered by module structure, and the second one is that context and shape are of equal importance to the semantic formation of oracle characters. The last result is that modules have strong local topology of the network.

The relationship between the local topology of molecular entanglements at the nanosized scale and the mechanical properties at the macroscopic level is one of the open questions in polymer science. In this work, the process of pore formation in amorphous polyethylene during uniaxial stretching is modeled within the framework of the molecular dynamics method. It is considered how a change in the concentration of intermolecular entanglements affects the dynamics of the formation of nanosized pores. The results of the statistical analysis of pore sizes indicate that intermolecular entanglements slow down the processes of their growth and aggregation in the bulk of the polymer.

Active distribution network (ADN) technology, as an important trend of the future smart distribution grid, is able to effectively absorb distributed energy resource (DER), to reasonably optimise grid‐load operation characteristics, and to safely support the reliability of power supply. Through enhancing energy utilisation efficiency and friendly interaction with user access, ADN technology is also able to comprehensively improve the power supply reliability of the distribution network. However, distributed feeder automation (FA), as an important part of ADN technology, will also meet new problems and challenges with the access of DER in the distribution network. The formal method can analyse the correctness and effectiveness of a distributed fault processing algorithm from mathematical logic, which provides an important theoretical basis for distributed fault processing. The focus herein is on the formal description and verification of topology modelling in fault location, isolation, and service restoration (FLISR) based on distributed processing. By abstracting and simplifying the complex power system features, the adaptability of the formal method is solved. The logical correctness of the topology model in FLISR based on distributed processing is verified. Finally, the distributed local topology model and algorithm is verified through a formal method using an actual ADN example.

Whole-brain network analysis is commonly used to investigate the topology of the brain in a variety of neuroimaging modalities. This approach is notable for its applicability to a large number of domains, such as understanding how brain network organization relates to cognition and behavior, examining disrupted brain network organization in disease, and assessing developmental trajectories across the lifespan. A benefit to this approach is the ability to summarize overall brain network organization with a single number (e.g., global efficiency). However, important local differences in network structure might exist without any corresponding observable differences in overall topology, making a whole-brain analysis strategy unlikely to detect relevant local findings. Here, we propose the network-based statistic (NBS) jackknife as a means of combining the utility of global network analysis strategies with the ability to detect relevant local differences in network structure. We describe the NBS jackknife framework, and demonstrate three specific testing scenarios in a series of examples. Finally, we provide an empirical example comparing global efficiency between children with ADHD and typically developing (TD) children. We demonstrate using functional connectivity data that there are no group differences in whole-brain global efficiency. Using the NBS jackknife, however, we identify group differences in global efficiency specific to the salience and subcortical subnetworks. The NBS jackknife framework has been implemented in a public, open source R package, netjack, available at https://cran.r-project.org/package=netjack.

The community detection problem has attracted a huge number of researchers from the scientific community because it helps to understand the relationship between structural and functional properties of social networks. In this paper, we propose a novel community detection approach namely, Closeness Similarity driven Information Diffusion based community detection (CSID) which makes full utilization of local topology for similarity computation followed by information exchange and modularity maximization. Most of the existing algorithms so far are not suitable to quantify both the local topology information and exchange of information between the nodes. Extensive experiments are carried out to perform a comparative evaluation on several real-world datasets. The experimental results show that the proposed approach gives better results in comparison to most of the baseline algorithms.

Cellular networks in a broad variety of materials influence many materials properties—a particularly important subject when it comes to microstructural coarsening of these networks. Hence, changes in the networks are of intense scientific and technological interest, where all topological changes occur in form of topological transitions. In two dimensions, there are three types of transitions, namely grain disappearance events, neighbour switching events, and edge disappearance events. Each event changes not only the morphology of a grain but also the local topology around this grain. The present work is the first investigation aimed at the question whether such topological transitions describe a topological random walk for individual grains or if they are a pure geometric necessity for structural space filling.

1AbstractProperties of city-level commuting networks are expected to influence epidemic potential of cities and modify the speed and spatial trajectory of epidemics when they occur. In this study, we use aggregated mobile phone user data to reconstruct commuter mobility networks for Bangkok (Thailand) and Dhaka (Bangladesh), two megacities in Asia with populations of 16 and 21 million people, respectively. We model the dynamics of directly-transmitted infections (such as SARS-CoV2) propagating on these commuting networks, and find that differences in network structure between the two cities drive divergent predicted epidemic trajectories: the commuting network in Bangkok is composed of geographically-contiguous modular communities and epidemic dispersal is correlated with geographic distance between locations, whereas the network in Dhaka has less distinct geographic structure and epidemic dispersal is less constrained by geographic distance. We also find that the predicted dynamics of epidemics vary depending on the local topology of the network around the origin of the outbreak. Measuring commuter mobility, and understanding how commuting networks shape epidemic dynamics at the city level, can support surveillance and preparedness efforts in large cities at risk for emerging or imported epidemics.

Novel topological methods are introduced to protein research. The aim is to identify hot-spot sites where a bifurcation can change the local topology of the protein backbone. Since the shape of a protein is intimately related to its biological function, a mutation that takes place at such a bifurcation hot-spot has an enhanced capacity to change the protein’s biological function. The methodology applies to any protein but it is developed with the SARS-CoV-2 spike protein as a timely example. First, topological criteria are introduced to identify and classify potential mutation hot-spot sites along the protein backbone. Then, the expected outcome of a substitution mutation is estimated for a general class of hot-spots, by a comparative analysis of the backbone segments that surround the hot-spot sites. This analysis employs the statistics of commensurable amino acid fragments in the Protein Data Bank, in combination with general stereochemical considerations. It is observed that the notorious D614G substitution of the spike protein is a good example of such a mutation hot-spot. Several topologically similar examples are then analyzed in detail, some of them are even better candidates for a mutation hot-spot than D614G. The local topology of the recently observed N501Y mutation is also inspected, and it is found that this site is prone to a different kind of local topology changing bifurcation.

Modelling air movement in sewer networks is needed in order to address the issues related to sewer odour complaints and sewer corrosions due to hydrogen sulphide in sewers. Most of the existing air flow models can only be applied in small sewer networks or the trunk lines of sewer systems. The purpose of this paper is therefore to propose a theoretical approach to formulate a general governing equation set for modelling steady air movement in large sewer systems. This approach decomposes the sewer system of interest into its basic physical components as pipes and nodes, and builds local topology of each pipe and each node based on geographic information system data as the fundamentals of model formulation. It avoids manually identifying each branch of the sewer system, eliminates the effect of physically closed networks in sewer systems on the governing equations, and considers key sewer components and all known driving forces. The proposed approach was applied to a real sewer system with over 500 pipes. The results show that the proposed model is applicable in modelling air movement in a large sewer system and provides a general idea of sewer gases moving through the system and their emission.

Dimension reduction is often used for several procedures of analysis of high dimensional biomedical data-sets such as classification or outlier detection. To improve the performance of such data-mining steps, preserving both distance information and local topology among data-points could be more useful than giving priority to visualization in low dimension. Therefore, we introduce topology-preserving distance scaling (TPDS) to augment a dimension reduction method meant to reproduce distance information in a higher dimension. Our approach involves distance inflation to preserve local topology to avoid collapse during distance preservation-based optimization. Applying TPDS on diverse biomedical data-sets revealed that besides providing better visualization than typical distance preserving methods, TPDS leads to better classification of data points in reduced dimension. For data-sets with outliers, the approach of TPDS also proves to be useful, even for purely distance-preserving method for achieving better convergence.

This paper presents a flexible free-form deformation (FFD) method that overcomes two limitations of many FFD methods: the cumbersome interface and the fixed topology of the control lattice. The novelty of the method mainly lies in two aspects: (1) a specifically-tailored T-spline volume with its local refinement is introduced to define the deformation space and construct the deformation tool, and (2) a proxy represented as a three-dimensional mesh close to the object to be deformed is constructed from the control lattice of the T-spline volume and is used to manipulate and drive the deformation. While the method inherits all the virtues of FFD such as C2 continuous local and global modifications, the proxy provides a much concise and intuitive interface for the user to edit the deformation in a way similar to cage-based deformations. Moreover, the method allows local topology changes of the proxy after parameterization, which is not available for most FFD methods or even cage-based deformations. These features enable more complicated free-form deformation to be achieved in a flexible and intuitive manner, as demonstrated by many experimental examples. Ministry of Education (MOE) This work was supported by the Ministry of Education, Singapore, under its MoE Tier-2 Grant (MoE 2017-T2-1- 076).

In this paper, we investigated the public goods game (PGG) on evolving random networks by introducing informed cooperators (IC) who can adjust their local topology of networks according to the information they obtained at cost. Our research shows that a high rate of network updating may provide a favorable environment for cooperation to emerge and thrive. In particular, we have found that the informed cooperators may separate the defectors (D) out and form an isolated population of cooperative strategies under adverse conditions.

Coupled neural P (CNP) systems are recently developed distributed and parallel computing models that are abstracted by the mechanisms of coupled and spiking neurons. CNP systems differ from spiking neural P (SNP) systems in two main ways, namely the utilization of three data units, and a coupled firing and dynamic threshold mechanism for neurons. This paper focuses on the application of CNP systems to solve multi-focus image fusion problems, and proposes a novel image fusion approach based on CNP systems. Based on two CNP systems with local topology, a multi-focus image fusion framework in the non-subsampled contourlet transform (NSCT) domain is developed, where the two CNP systems are utilized to control the fusion of low-frequency coefficients in the NSCT domain. The proposed fusion approach is evaluated on an open data set of 19 multi-focus images based on five fusion quality indices, and compared to 11 state-of-the-art fusion approaches. Quantitative and qualitative experimental results demonstrate the advantages of the proposed fusion approach in terms of visual quality and fusion performance.

Properties of city-level commuting networks are expected to influence epidemic potential of cities and modify the speed and spatial trajectory of epidemics when they occur. In this study, we use aggregated mobile phone user data to reconstruct commuter mobility networks for Bangkok (Thailand) and Dhaka (Bangladesh), two megacities in Asia with populations of 16 and 21 million people, respectively. We model the dynamics of directly-transmitted infections (such as SARS-CoV-2) propagating on these commuting networks, and find that differences in network structure between the two cities drive divergent predicted epidemic trajectories: the commuting network in Bangkok is composed of geographically-contiguous modular communities and epidemic dispersal is correlated with geographic distance between locations, whereas the network in Dhaka has less distinct geographic structure and epidemic dispersal is less constrained by geographic distance. We also find that the predicted dynamics of epidemics vary depending on the local topology of the network around the origin of the outbreak. Measuring commuter mobility, and understanding how commuting networks shape epidemic dynamics at the city level, can support surveillance and preparedness efforts in large cities at risk for emerging or imported epidemics.

This paper proposes a local topology correction method for large power grid simulation. The analysis method of the local topology corresponding to the “open” and “close” events of the power equipment and the correction method of the corresponding node admittance matrix. This method is suitable for online analysis systems such as large power grid simulation systems. Finally, this paper verifies that the local topology correction method proposed in this paper greatly improves the topology analysis speed and calculation efficiency of the large power grid training simulation system.

It has been shown that an attacker can stealthily launch false data injection attacks against the state estimation without knowing the full topology and parameter information of the entire power network. In this paper, we propose a heuristic yet effective method to determine a feasible attacking region of a single line, which requires less network information. We use six IEEE standard systems to validate the proposed attacking strategy. This paper can reveal the vulnerability of the real-time topology of a power grid and is very helpful to develop effective protection strategies against topology attacks in smart grids.

The von Neumann-Mullins-relation is one of the most fundamental laws of microstructural coarsening enabling a prediction of the kinetics of individual cells in many cellular structures ranging from foams to polycrystalline metals and alloys. The derivation of this relation is based on the assumption of ideal coarsening, regular cells, and a constant environment. However, in reality this is rarely the case. Hence, in the present work it is shown how the local topology and local topological changes influence the growth kinetics of individual grains.