Jiejun Xu

Bio: Jiejun Xu is an academic researcher from HRL Laboratories. The author has contributed to research in topics: Graph (abstract data type) & Social media. The author has an hindex of 18, co-authored 72 publications receiving 920 citations. Previous affiliations of Jiejun Xu include University of California, Santa Barbara & General Motors.

Graph convolutional networks: a comprehensive review

Abstract: Graphs naturally appear in numerous application domains, ranging from social analysis, bioinformatics to computer vision. The unique capability of graphs enables capturing the structural relations among data, and thus allows to harvest more insights compared to analyzing data in isolation. However, it is often very challenging to solve the learning problems on graphs, because (1) many types of data are not originally structured as graphs, such as images and text data, and (2) for graph-structured data, the underlying connectivity patterns are often complex and diverse. On the other hand, the representation learning has achieved great successes in many areas. Thereby, a potential solution is to learn the representation of graphs in a low-dimensional Euclidean space, such that the graph properties can be preserved. Although tremendous efforts have been made to address the graph representation learning problem, many of them still suffer from their shallow learning mechanisms. Deep learning models on graphs (e.g., graph neural networks) have recently emerged in machine learning and other related areas, and demonstrated the superior performance in various problems. In this survey, despite numerous types of graph neural networks, we conduct a comprehensive review specifically on the emerging field of graph convolutional networks, which is one of the most prominent graph deep learning models. First, we group the existing graph convolutional network models into two categories based on the types of convolutions and highlight some graph convolutional network models in details. Then, we categorize different graph convolutional networks according to the areas of their applications. Finally, we present several open challenges in this area and discuss potential directions for future research.

DEMO-Net: Degree-specific Graph Neural Networks for Node and Graph Classification

Abstract: Graph data widely exist in many high-impact applications. Inspired by the success of deep learning in grid-structured data, graph neural network models have been proposed to learn powerful node-level or graph-level representation. However, most of the existing graph neural networks suffer from the following limitations: (1) there is limited analysis regarding the graph convolution properties, such as seed-oriented, degree-aware and order-free; (2) the node's degreespecific graph structure is not explicitly expressed in graph convolution for distinguishing structure-aware node neighborhoods; (3) the theoretical explanation regarding the graph-level pooling schemes is unclear. To address these problems, we propose a generic degree-specific graph neural network named DEMO-Net motivated by Weisfeiler-Lehman graph isomorphism test that recursively identifies 1-hop neighborhood structures. In order to explicitly capture the graph topology integrated with node attributes, we argue that graph convolution should have three properties: seed-oriented, degree-aware, order-free. To this end, we propose multi-task graph convolution where each task represents node representation learning for nodes with a specific degree value, thus leading to preserving the degreespecific graph structure. In particular, we design two multi-task learning methods: degree-specific weight and hashing functions for graph convolution. In addition, we propose a novel graph-level pooling/readout scheme for learning graph representation provably lying in a degree-specific Hilbert kernel space. The experimental results on several node and graph classification benchmark data sets demonstrate the effectiveness and efficiency of our proposed DEMO-Net over state-of-the-art graph neural network models.

Civil Unrest Prediction: A Tumblr-Based Exploration

Abstract: This work focuses on detecting emerging civil unrest events by analyzing massive micro-blogging streams. Specifically, we propose an early detection system consisting of a novel cascade of text-based filters to identify civil unrest event posts based on their topics, times and locations. In contrast to the model-based prediction approaches, our method is purely extractive as it detects relevant posts from massive volumes of data directly. We design and implement such a system in a distributed framework for scalable processing of real world data streams. Subsequently, a large-scale experiment is carried out on our system with the entire dataset from Tumblr for three consecutive months. Experimental result indicates that the simple filter-based method provides an efficient and effective way to identify posts related to real world civil unrest events. While similar tasks have been investigated in different social media platforms (e.g., Twitter), little work has been done for Tumblr despite its popularity. Our analysis on the data also shed light on the collective micr-oblogging patterns of Tumblr.

Rolling through tumblr: characterizing behavioral patterns of the microblogging platform

Abstract: Tumblr, a microblogging platform and social media website, has been gaining popularity over the past few years. Despite its success, little has been studied on the human behavior and interaction on this platform. This is important as it sheds light on the driving force behind Tumblr's growth. In this work, we present a quantitative study of Tumblr based on the complete data coverage for four consecutive months consisting of 23.2 million users and 10.2 billion posts. We first explore various attributes of users, posts, and tags in detail and extract behavioral patterns based on the user generated content. We then construct a massive reblog network based on the primary user interactions on Tumblr and present findings on analyzing its topological structure and properties. Finally, we show substantial results on providing location-specific usage patterns from Tumblr, despite no built-in support for geo-tagging or user location functionality. Essentially this is done by conducting a large-scale user alignment with a different social media platform (e.g., Twitter) and subsequently propagating geo-information across platforms. To the best of our knowledge, this work is the first attempt to carry out large-scale measurement-driven analysis on Tumblr.

Graph Convolutional Networks: Algorithms, Applications and Open Challenges

Abstract: Graph-structured data naturally appear in numerous application domains, ranging from social analysis, bioinformatics to computer vision. The unique capability of graphs enables capturing the structural relations among data, and thus allows to harvest more insights compared to analyzing data in isolation. However, graph mining is a challenging task due to the underlying complex and diverse connectivity patterns. A potential solution is to learn the representation of a graph in a low-dimensional Euclidean space via embedding techniques that preserve the graph properties. Although tremendous efforts have been made to address the graph representation learning problem, many of them still suffer from their shallow learning mechanisms. On the other hand, deep learning models on graphs have recently emerged in both machine learning and data mining areas and demonstrated superior performance for various problems. In this survey, we conduct a comprehensive review specifically on the emerging field of graph convolutional networks, which is one of the most prominent graph deep learning models. We first introduce two taxonomies to group the existing works based on the types of convolutions and the areas of applications, then highlight some graph convolutional network models in details. Finally, we present several challenges in this area and discuss potential directions for future research.

Random graphs

Abstract: We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Graph Neural Networks: A Review of Methods and Applications

Abstract: Lots of learning tasks require dealing with graph data which contains rich relation information among elements. Modeling physics systems, learning molecular fingerprints, predicting protein interface, and classifying diseases demand a model to learn from graph inputs. In other domains such as learning from non-structural data like texts and images, reasoning on extracted structures (like the dependency trees of sentences and the scene graphs of images) is an important research topic which also needs graph reasoning models. Graph neural networks (GNNs) are neural models that capture the dependence of graphs via message passing between the nodes of graphs. In recent years, variants of GNNs such as graph convolutional network (GCN), graph attention network (GAT), graph recurrent network (GRN) have demonstrated ground-breaking performances on many deep learning tasks. In this survey, we propose a general design pipeline for GNN models and discuss the variants of each component, systematically categorize the applications, and propose four open problems for future research.