The different factors involved in the growth process of complex networks imprint valuable information in their observable topologies. How to exploit this information to accurately predict structural network changes is the subject of active research. A recent model of network growth sustains that the emergence of properties common to most complex systems is the result of certain trade-offs betwe...

We present a novel approach to embedding data represented by a network into a lowdimensional Euclidean space. Unlike existing methods, the proposed method attempts to minimize an energy function based on the cross-entropy between desirable and embedded node configurations without directly utilizing pairwise distances between nodes. We also propose a natural criterion to effectively evaluate an ...

Node classification in structural networks has been proven to be useful many real world applications. With the development of network embedding, performance node greatly improved. However, nearly all existing embedding based methods are hard capture actual category features a because linearly inseparable problem low-dimensional space; meanwhile they cannot incorporate simultaneously structure i...

Traditional stock movement prediction tasks are formulated as either classification or regression task, and the relation between stocks not considered an input of prediction. The relative order ranking is more important than price return a single for making proper investment decisions. Stock performance can be improved by incorporating information in task. We employ graph-based approach use mac...

Network embedding aspires to learn a low-dimensional vector of each node in networks, which can apply diverse data mining tasks. In real-life, many networks include rich attributes and temporal information. However, most existing approaches ignore either information or network attributes. A self-attention based architecture using higher-order weights for both static attributed is presented this...

A widely established set of unsupervised node embedding methods can be interpreted as consisting two distinctive steps: i) the definition a similarity matrix based on graph interest followed by ii) an explicit or implicit factorization such matrix. Inspired this viewpoint, we propose improvements in both steps framework. On one hand, to encode similarities free energy distance, which interpolat...

This paper investigates the problem of network embedding, which aims at learning low-dimensional vector representation of nodes in networks. Most existing network embedding methods rely solely on the network structure, i.e., the linkage relationships between nodes, but ignore the rich content information associated with it, which is common in real world networks and beneficial to describing the...