Link prediction in networks has become a growing concern among researchers. In this paper, various link methods are compared for better results intercity transportation networks. For practical consideration, new index is proposed to sort the of different algorithms taking both predicting precision and number existing links into consideration. According index, we find best threshold determine existence with simulate anneal. Experiments show that performs well

Link prediction in social networks has become a growing concern among researchers. In this paper, the clustering method was used to exploit grouping tendency of nodes, and index (CI) proposed predict potential links with characteristics scientific cooperation network taken into consideration. Results showed that CI performed better than traditional indices for coauthorship by compensating their disadvantages. Compared algorithms, specific type can reflect features achieve more accurate predictions.

The research of hypersonic gliding lifting body vehicle is so complicated that its aerodynamic characteristics prediction very time-consuming. To solve the problem, parametric design method based on class function/shape function transformation and power employed, a shape with five configuration parameters which convenient for analysis proposed. At same time, we calculate panel element kinds engineering-estimation methods according to compression or expansion elements. Compared numerical...

Link prediction in social networks has become a growing concern among researchers. In this paper, clustering index (CI) is calculated to predict potential links with characteristics of scientific cooperation network taken into consideration. Compared traditional universal algorithms, algorithm for specific type can better reflect the features and achieve predicting results. Experiments show that CI performs than indices coauthorship networks.

Reinforcement learning, as an effective method to solve complex sequential decision-making problems, plays important role in areas such intelligent and behavioral cognition. It is well known that the sample experience replay mechanism contributes development of current deep reinforcement learning by reusing past samples improve efficiency samples. However, existing priority changes distribution set due higher sampling frequency assigned a specific transition, it cannot be applied...

The successful application of deep reinforcement learning in RTS games such as StarCraft II has inspired people to apply multi-agent learning(MADRL) more fields. In the field wargame, hexagonal maps are often used for simulation, which can't adapt rapid development wargame. continuous space we construct a ship-defense scenario that includes multiple aircraft and ships. We Q network(DQN) method MADRL, CNN extract features entities, centralized distributed decision-making training architecture...

An organization is a collection of functional members for specific purpose and their relationships. In reality, various tangible intangible organizations affect all aspects people's lives. This paper proposes F-statistics group division to evaluate define the difference degrees groups. The results can be used as an indicator number estimation, verify effectiveness based estimation method.

Countermeasures of safety improvements for rural highways are great significance to crash reduction and enhancement. In order compare prioritize the effects different improvement measures with certain quantitative indexes in advance actual implementation, a theoretically-synthetic evaluation model traffic was built based on combination analytic hierarchy process (AHP) fuzzy logic theory. A MATLAB program also developed improve assessment efficiency. The experts' scoring method employed...

Network embedding technology transforms network structure into node vectors, which reduces the complexity of representation and can be effectively applied to tasks such as classification, reconstruction link prediction. The most important concern is how retain local structural features while capturing global network. In view shortcomings SDNE in weighted directed network, this paper proposes an improved model based on degree. experimental results show that has better effect than original...

Heterogeneous graph representation learning aims to learn meaningful vectors from heterogeneous networks in low dimension, so as realize the extraction of structure and attribute features networks. Embedding vector is basis key complex network analysis, which can be used downstream tasks. The points neural are: how define neighbors aggregate them. Although a lot work has been devoted homogeneous or representation, effective combination information node information, especially use meta-path...

Link prediction is an important application in complex networks. It predicts existing but undiscovered associations or possible future relationships the network. However, networks real life have much noise. The we observe are incomplete redundant which interfere with effect of link prediction. This paper summarizes and constructs four kinds common noises social networks, then analyzes robustness traditional methods based on network representation under influence different degrees multiple...