Recent Research Projects

NeuGraph (NGra): System for Graph Neural Networks (GNNs)

Recent deep learning models have moved beyond low-dimensional regular grids to high-dimensional graph-structured data, leading to large graph-based irregular and sparse models that go beyond what existing DL frameworks are designed for. We introduce NeuGraph (NGra), a parallel processing framework for graph-based deep neural networks (GNNs) on top of existing DL frameworks.

  • NeuGraph presents a new programming model for expressing deep neural networks as vertex programs. This model not only allows GNNs to be expressed intuitively, but also facilitates the mapping to an efficient dataflow representation.
  • NeuGraph addresses the scalability challenge transparently through automatic graph partitioning and chunk-based stream processing out of GPU core or over multiple GPUs.
  • NeuGraph achieves efficiency through highly optimized Scatter/Gather operations on GPU despite graph sparsity.

Garaph: GPU-accelerated Graph Processing System

Recent advances in storage and accelerators provide the opportunity to efficiently process large-scale graphs on a single machine. Thus, we present Garaph, a GPU-accelerated graph processing system. Garaph is novel in three ways:

  • First, Garaph proposes a vertex replication degree customization scheme that maximizes the GPU utilization given vertices’ degrees and space constraints.
  • Second, Garaph adopts a balanced edge-based partition method, ensuring sequential memory access and load balance over CPU threads, and also a hybrid of notify-pull and pull computation models optimized for fast graph processing on the CPU.
  • Third, Garaph designs a workload scheduler which takes into account both characteristics of processing elements and graph algorithms.

Publications

2019

Autor3: Automated Real-time Ranking with Reinforcement Learning in Ecommerce Sponsored Search Advertising
Yusi Zhang, Zhi Yang*, Liang Wang, and Li He
The 28th ACM International Conference on Information and Knowledge Management (CIKM), 2019

ScheD2: Scheduling Deep Learning Training via Deep Reinforcement Learning
Yunteng Luan, Xukun Chen, Hanyu Zhao, Zhi Yang*, Yafei Dai
IEEE Global Communications Conference (GlobeCom), 2019

NeuGraph: Parallel Deep Neural Network Computation on Large Graphs
Lingxiao Ma, Zhi Yang*, Youshan Miao, Jilong Xue, Ming Wu, Lidong Zhou, Yafei Dai
USENIX Annual Technical Conference (USENIX ATC), 2019

SeerNet: Predicting Convolutional Neural Network Feature-Map Sparsity through Low-Bit Quantization
Shijie Cao, Lingxiao Ma, Wencong Xiao, Chen Zhang, Yunxin Liu, Lintao Zhang, Lanshun Nie, Zhi Yang
30th IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019

2018

SDPaxos: Building Efficient Semi-Decentralized Geo-replicated State Machines
Hanyu Zhao, Quanlu Zhang, Zhi Yang*, Ming Wu and Yafei Dai
ACM Symposium on Cloud Computing 2018 (SoCC), 2018

A Game Theoretic Model For The Formation of Navigable Small-world Networks--The Tradeoff Between Distance And Reciprocity
Zhi Yang, Wei Chen
to be appeared in ACM Transactions on Internet Technology (TOIT), 2018

DeepCredit: Clickstream-based model for loan Risk Prediction in P2P Lending
Zhi Yang, Yusi Zhang, Binghui Guo, Ben Zhao and Yafei Dai
International AAAI Conference on Web and Social Media (ICWSM), 2018

Defending against Social Network Sybils with Interaction Graph Embedding
Zhi Yang, Yusi Zhang and Yafei Dai
International IEEE Conference on Communications and Network Security (CNS), 2018

Proactive Video Push for Optimizing Bandwidth Consumption in Hybrid CDN-P2P VoD Systems
Yuanxing Zhang, Chengliang Gao, Yangze Guo, Kaigui Bian, Xin Jin, Zhi Yang, Lingyang Song, Jiangang Cheng, Hu Tuo, and Xiaoming Li
IEEE International Conference on Computer Communications (INFOCOM), 2018

ATDPS: An Adaptive Time-Dependent Push Strategy in Hybrid CDN-P2P VoD System
Yangze Guo, Yuanxing Zhang, Zhi Yang*, Kaigui Bian, Hu Tuo and Yafei Dai
International In proceedings of IEEE ICC (ICC), 2018

2017

Garaph: Efficient GPU-accelerated Graph Processing on a Single Machine with Balanced Replication
Lingxiao Ma, Zhi Yang*, Han Chen, Jilong Xue, Yafei Dai
USENIX Annual Technical Conference (USENIX ATC), 2017

BCStore: Bandwidth-Efficient In-memory KV-Store with Batch Coding
Shenglong Li, Quanlu Zhang, Zhi Yang*, Yafei Dai
International Conference on Massive Storage Systems and Technology (MSST), 2017

DeltaCFS: Boosting Delta Sync for Cloud Storage Services by Learning from NFS
Quanlu Zhang, Zhenhua Li, Zhi Yang, Shenglong Li, Yangze Guo, Yafei Dai
IEEE International Conference on Distributed Computing Systems (ICDCS), 2017

2016

Efficient Distributed Machine Learning with Trigger Driven Parallel Training
Shenglong Li,Jilong Xue, Zhi Yang, Yafei Dai
IEEE Global Communications Conference (GlobeCom), 2016

Processing Concurrent Graph Analytics with Decoupled Computation Model
Jilong Xue, Zhi Yang*, Shian Hou, Yafei Dai
IEEE Transaction On Computer (TOC), 2016

2015

VoteTrust: Leveraging Friend Invitation Graph to Defend against Social Network Sybils
Zhi Yang, Jilong Xue, Xiaoyong Yang, Xiao Wang, Yafei Dai
IEEE Transactions on Dependable and Secure Computing (TDSC), 2015

Understanding and Surpassing Dropbox: Efficient Incremental Synchronization in Cloud Storage Services
Shenglong Li, Quanlu Zhang, Zhi Yang, and Yafei Dai
IEEE Global Communications Conference (GlobeCom), 2015

When Graph Computation Meets Heterogeneous Cluster
Jilong Xue, Zhi Yang*, Shian Hou, Yafei Dai
IEEE International Conference on Big Data (BigData), 2015

Process-driven Analysis of Dynamics in Online Social Interactions
Zhi Yang, Jilong Xue, Christo Wilson, Ben Y. Zhao, and Yafei Dai
ACM Conference on Online Social Networks (COSN), 2015

A game theoretic model for the formation of navigable small-world networks
Zhi Yang and Wei Chen
International World Wide Web Conference (WWW), 2015

CHARM: A Cost-efficient Multi-cloud Data Hosting Scheme with High Availability
Quanlu Zhang, Shenglong Li, Zhenhua Li, Yuanjian Xing, Zhi Yang, and Yafei Dai
IEEE Transaction on Cloud Computing (TCC), 2015

Uncovering User Interaction Dynamics in Online Social Networks
Zhi Yang, Jilong Xue, Christo Wilson, Ben Y. Zhao, and Yafei Dai
International AAAI Conference on Weblogs and Social Media (ICWSM), 2015

2014

Uncovering Social Network Sybils in the Wild
Zhi Yang , Christo Wilson, Xiao Wang, Tingting Gao, Ben Y. Zhao and Yafei Dai
ACM Transactions on Knowledge Discovery from Data (TKDD), 2014

Seraph: an Efficient, Low-cost System for Concurrent Graph Processing
Jilong Xue, Zhi Yang *, Zhi Qu, Shian Hou, Yafei Dai
23nd International ACM Symposium on High-Performance Parallel and Distributed Computing (HPDC), 2014

Understanding the Performance of Offline Download in Real P2P Networks
Zhi Yang ,Yuanjian Xing,Chi Chen,Yafei Dai
Int'J of Peer-to-Peer Networking and Applications, 2014

Before 2013

Seraph: An Efficient System for Parallel Processing on a Shared Graph
Zhi Yang , Jilong Xue, Zhi Qu, Shian Hou, Yafei Dai
7th Workshop on Large-Scale Distributed Systems and Middleware (LADIS), 2013

VoteTrust: Leveraging Friend Invitation Graph to Defend against Social Network Sybils
Jilong Xue, Zhi Yang*, Xiaoyong Yang, Xiao Wang, Lijiang Chen, Yafei Dai
32nd Annual IEEE International Conference on Computer Communications (INFOCOM), 2013

PROB: a Lightweight Approach for Fast Content Propagation in P2P Networks
Zhi Yang and Yafei Dai
IEEE Thirteen International Conference on Peer-to-Peer Computing (P2P), 2013

Beehive: Low-cost Content Subscription Service using Cloudlets
YuanJian Xing, Zhi Yang*, Chi Chen, Yafei Dai
SCIENCE CHINA Information Sciences, 2013

Uncovering Social Network Sybils in the Wild
Zhi Yang , Christo Wilson, Xiao Wang, Tingting Gao, Ben Y. Zhao and Yafei Dai
11th ACM SIGCOMM Internet Measurement Conference (IMC), 2011

Protector: a Probabilistic Failure Detector for Cost-effective Peer-to-Peer Storage,
Zhi Yang , Jing Tian, Ben Y. Zhao, Wei Chen, Yafei Dai
IEEE Transactions on Parallel and Distributed Systems (TPDS), 2011

On the QoS of Offline Download in Retrieving Peer-side File Resource
Yuanjian Xing, Zhi Yang*, Chi Chen, Jilong Xue and Yafei Dai
40th International Conference on Parallel Processing (ICPP), 2011

Exploring Peer Heterogeneity: Towards Understanding and Application
Zhi Yang , Yuanjian Xing, Feng Xiao, Zhi Qu, Xiaoming Li and Yafei Dai
11th International Conference on Peer-to-Peer Computing(P2P), 2011

AutoProc: An Automatic Proactive Replication Scheme for P2P Storage
Zhi Yang and Yafei Dai
SCIENCE CHINA Information Sciences, 2011

AmazingStore: Available, Low-cost Online Storage Service Using Cloudlets
Zhi Yang , Ben Y. Zhao, Yuanjian Xing, Song Ding, Feng Xiao and Yafei Dai
9th International Workshop on P2P Systems (IPTPS), 2010

Towards a More Accurate Availability Evaluation in Peer-to-Peer Storage Systems
Zhi Yang, Jing Tian and Yafei Dai.
Int. J. High Performance Computing and Networking, 2010

Exploring cost-availability tradeoff in P2P storage system
Zhi Yang, Yafei Dai and Xiao Zhen.
38th International Conference on Parallel Processing(ICPP), 2009

The Neutralizer: a Self-Configurable Failure Detector for Minimizing Distributed Storage Maintenance Cost
Zhi Yang, Yafei Dai and Xiaoming Li..
In Concurrency and Computation: Practice and Experience, 2009

Exploring cost-availability tradeoff in P2P storage system
Jing Tian, Zhi Yang, Wei Chen, Ben Y. Zhao, Yafei Dai.
27th International Symposium on Reliable Distributed Systems(SRDS), 2008

A Data Placement Scheme with Time-Related Model for P2P Storages
Jing Tian, Zhi Yang and Yafei Dai
7th International Conference on Peer-to-Peer Computing(P2P), 2007

Towards a More Accurate Availability Evaluation in Peer-to-Peer Storage Systems
Zhi Yang, Jing Tian and Yafei Dai
IEEE International Conference on Networking, Architecture and Storage(NAS), 2006

Services

PC member

ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) 2019
International Conference Series on Advances in Social Network Analysis and Mining (ASONAM) 2019
IEEE International Conference on Big Data (IEEE BigData) 2018, 2019
The Web Conference (WWW) 2017 (Poster Track)