TY - GEN
T1 - Discerning edge influence for network embedding
AU - Wang, Yaojing
AU - Yao, Yuan
AU - Tong, Hanghang
AU - Xu, Feng
AU - Lu, Jian
N1 - Funding Information:
This work is supported by the National Natural Science Foundation of China (No. 61632021, 61672274, 61702252) and the Collaborative Innovation Center of Novel Software Technology and Industrialization. Hanghang Tong is partially supported by NSF (IIS-1651203, IIS-1715385), and DHS (2017-ST-061-QA0001). Yuan Yao is the corresponding author.
Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019/11/3
Y1 - 2019/11/3
N2 - Network embedding, which learns the low-dimensional representations of nodes, has gained significant research attention. Despite its superior empirical success, often measured by the prediction performance of downstream tasks (e.g., multi-label classification), it is unclear why a given embedding algorithm outputs the specific node representations, and how the resulting node representations relate to the structure of the input network. In this paper, we propose to discern the edge influence as the first step towards understanding skip-gram basd network embedding methods. For this purpose, we propose an auditing framework Near, whose key part includes two algorithms (Near-add and Near-del) to effectively and efficiently quantify the influence of each edge. Based on the algorithms, we further identify high-influential edges by exploiting the linkage between edge influence and the network structure. Experimental results demonstrate that the proposed algorithms (Near-add and Near-del) are significantly faster (up to 2, 000×) than straightforward methods with little quality loss. Moreover, the proposed framework can efficiently identify the most influential edges for network embedding in the context of downstream prediction task and adversarial attacking.
AB - Network embedding, which learns the low-dimensional representations of nodes, has gained significant research attention. Despite its superior empirical success, often measured by the prediction performance of downstream tasks (e.g., multi-label classification), it is unclear why a given embedding algorithm outputs the specific node representations, and how the resulting node representations relate to the structure of the input network. In this paper, we propose to discern the edge influence as the first step towards understanding skip-gram basd network embedding methods. For this purpose, we propose an auditing framework Near, whose key part includes two algorithms (Near-add and Near-del) to effectively and efficiently quantify the influence of each edge. Based on the algorithms, we further identify high-influential edges by exploiting the linkage between edge influence and the network structure. Experimental results demonstrate that the proposed algorithms (Near-add and Near-del) are significantly faster (up to 2, 000×) than straightforward methods with little quality loss. Moreover, the proposed framework can efficiently identify the most influential edges for network embedding in the context of downstream prediction task and adversarial attacking.
KW - Edge Influence
KW - Network Embedding
KW - Network Topological Properties
UR - http://www.scopus.com/inward/record.url?scp=85075436437&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85075436437&partnerID=8YFLogxK
U2 - 10.1145/3357384.3358044
DO - 10.1145/3357384.3358044
M3 - Conference contribution
AN - SCOPUS:85075436437
T3 - International Conference on Information and Knowledge Management, Proceedings
SP - 429
EP - 438
BT - CIKM 2019 - Proceedings of the 28th ACM International Conference on Information and Knowledge Management
PB - Association for Computing Machinery
T2 - 28th ACM International Conference on Information and Knowledge Management, CIKM 2019
Y2 - 3 November 2019 through 7 November 2019
ER -