An adaptive greedy flow routing algorithm for performance improvement in software‐defined network

Recently, there has been a surge of the video services over the Internet. However, service providers still have difficulties in providing high-quality video streaming due to the problem of scheduling efficiency and the wide fluctuations of end-to-end delays in the existing multi-path algorithms. To solve these two problems affecting video transmission quality, networks are expected to have the capability of dynamically managing the network nodes for satisfying quality-of-service requirements, which is a challenging issue for media streaming applications. Against this changing network landscape, this article proposes a dynamic and adaptive multi-path routing algorithm under three constraints (packet loss, time delay, and bandwidth) that are based on software-defined network for centralized routing computations and real-time network state updating in multimedia applications. Compared with related multi-path routing proposals, dynamic and adaptive multi-path routing makes efficient use of the latest global network state information achieved by the OpenFlow controller and calculates the optimal routes dynamically according to the real-time status information of the link. Moreover, our proposed algorithm can significantly reduce the computational overhead of the controller while completing a fine-grained flow balance. Experimental results show that dynamic and adaptive multi-path routing significantly outperforms other existing scheduling approaches in achieving a 35%–70% improvement in quality-of-service.

Download Full-text

Multi-Labeled Relation Extraction with Attentive Capsule Network

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v33i01.33017484 ◽

2019 ◽

Vol 33 ◽

pp. 7484-7491 ◽

Cited By ~ 6

Author(s):

Xinsong Zhang ◽

Pengshuai Li ◽

Weijia Jia ◽

Hai Zhao

Keyword(s):

Performance Improvement ◽

Loss Function ◽

State Of The Art ◽

Routing Algorithm ◽

Relation Extraction ◽

Neural Models ◽

Novel Approach ◽

Art Works ◽

Significant Performance ◽

Better Than

To disclose overlapped multiple relations from a sentence still keeps challenging. Most current works in terms of neural models inconveniently assuming that each sentence is explicitly mapped to a relation label, cannot handle multiple relations properly as the overlapped features of the relations are either ignored or very difficult to identify. To tackle with the new issue, we propose a novel approach for multi-labeled relation extraction with capsule network which acts considerably better than current convolutional or recurrent net in identifying the highly overlapped relations within an individual sentence. To better cluster the features and precisely extract the relations, we further devise attention-based routing algorithm and sliding-margin loss function, and embed them into our capsule network. The experimental results show that the proposed approach can indeed extract the highly overlapped features and achieve significant performance improvement for relation extraction comparing to the state-of-the-art works.

Download Full-text