Jiwei Wang

Jiwei Wang

Auto-generated profile, based on publicly available information

Jiwei Wang

Hangzhou Dianzi University

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 October 2020 in IEEE Transactions on Cloud Computing
Verified
Journal Article
Closed access
CJCongfeng JIANG
Congfeng JIANG
YQYitao Qiu
Yitao Qiu
WSWeisong Shi
Weisong Shi
ZGZhefeng Ge
Zhefeng Ge
JWJiwei Wang
Jiwei Wang
JLJiangbin Lin
Jiangbin Lin
Workload characteristics are vital for both data center operation and job scheduling in co-located data centers, where online services and batch jobs are deployed on the same production cluster. In this paper, a comprehensive analysis is conducted on Alibaba‘s cluster-trace-v2018 of a production cluster of 4034 machines. The findings and insights are the following: (1) The workload on the production cluster poses a daily cyclical fluctuation, in terms of CPU and disk I/O utilization, and the memory system has become the performance bottleneck of a co-located cluster. (2) Batch jobs including their tasks and derived instances can be approximated as Zipf distribution. However, for all batch jobs with directed acyclic graph dependency, they suffer from co-location with online services since the online services are highly prioritized. (3) The resource usages of containers have similar cyclical fluctuation consistent with the whole cluster, while their memory usages remain approximately constant. (4) The number of batch jobs co-located with online services is dependent on the mispredictions per kilo instructions of online services. In order to guarantee the QoS of online services, when the MPKI of online services rises, the number of batch jobs to be co-located on the same machine should decrease.

1 November 2015
Verified
Conference Paper
Closed access
JWJiwei Wang
Jiwei Wang
FLFangai Liu
Fangai Liu
XLXu Li
Xu Li
HLHaoran Liu
Haoran Liu
XZXiaohui Zhao
Xiaohui Zhao
Currently, online social networks are experiencing explosive growth, and play important roles in all aspects of lives, such as daily communication, online study and online dating. People's everyday life can't stand away from online social networks. With the advances of online social networks and popularity of intelligent mobile devices, more and more service providers are marketing their own online social network services, which providing competent services and ample functionalities. Nonetheless, almost all the mainstream social networks are based on the centralized servers or distributed server clusters. The whole system functions on the basis of the servers, so users depend excessively on servers. This kind of system architecture has potential single-point or multi-point failures though it can offer superior services to users. Users can't log into the system to obtain social services when the servers run into faults, such as equipment failures, link failures or network attacks. More seriously, ifcentralized databases meet with troubles, physical damages or mistake operations, massive user data perhaps could not be used or even lost. This paper defines the problem as Local Service Fault Partition (LSFP) and proposes a novel social network model called HPOSN which applies thoughts of P2P to solve LSFP and optimizes the centralized social network services. The prototype of HPOSN optimizes the communication overhead to control the flooding problem of P2P according to social relationship. Several simulation experiments with two parameters have been conducted, including the Recovery Time and the Recovery Success Rate. Results indicate that HPOSN can solve LSFP problem in centralized social networks pragmatically. Users in the LSFP area can recovery the social services locally through selforganizing and self-managing when they lose the services from centralized server or distributed server clusters. So HPOSN can improve the stability of social network services and user experiences.

2 February 2021
Verified
Book Chapter
Closed access
PDF file
JWJiwei Wang
Jiwei Wang
YLYuegang Li
Yuegang Li
CJCongfeng Jiang
Congfeng Jiang
CMChao Ma
Chao Ma
LTLinlin Tang
Linlin Tang
SGShuangshuang Guo
Shuangshuang Guo

16 October 2020 in arXiv
Verified
Preprint
Open access
PDF file
SVSunny Verma
Sunny Verma
JWJiwei Wang
Jiwei Wang
ZGZhefeng Ge
Zhefeng Ge
RSRujia Shen
Rujia Shen
Jin FanJin Fan
Jin Fan
WLWei Liu
Wei Liu
Multimodal sentiment analysis utilizes multiple heterogeneous modalities for sentiment classification. The recent multimodal fusion schemes customize LSTMs to discover intra-modal dynamics and design sophisticated attention mechanisms to discover the inter-modal dynamics from multimodal sequences. Although powerful, these schemes completely rely on attention mechanisms which is problematic due to two major drawbacks 1) deceptive attention masks, and 2) training dynamics. Nevertheless, strenuous efforts are required to optimize hyperparameters of these consolidate architectures, in particular their custom-designed LSTMs constrained by attention schemes. In this research, we first propose a common network to discover both intra-modal and inter-modal dynamics by utilizing basic LSTMs and tensor based convolution networks. We then propose unique networks to encapsulate temporal-granularity among the modalities which is essential while extracting information within asynchronous sequences. We then integrate these two kinds of information via a fusion layer and call our novel multimodal fusion scheme as Deep-HOSeq (Deep network with higher order Common and Unique Sequence information). The proposed Deep-HOSeq efficiently discovers all-important information from multimodal sequences and the effectiveness of utilizing both types of information is empirically demonstrated on CMU-MOSEI and CMU-MOSI benchmark datasets. The source code of our proposed Deep-HOSeq is and available at https://github.com/sverma88/Deep-HOSeq--ICDM-2020.