Edge-Cloud Polarization and Collaboration: A Comprehensive Survey for AI

Yao, Jiangchao; Zhang, Shengyu; Yang, Yue; Wang, Feng; Ma, Jianxin; Zhang, Jianwei; Chu, Yunfei; Luo, Jing‐Li; Jia, Kunyang; Shen, Tao; Wu, Anpeng; Zhang, Fengda; Tan, Ziqi; Kuang, Kun; Wu, Chao; Wu, Fei; Zhou, Jingren; Yang, Hongxia

doi:10.1109/tkde.2022.3178211

Cited by 37 publications

(30 citation statements)

References 236 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For surrogate methods, there is no consensus on how to define neighbours of the input graph data and how to choose interpretable surrogate models [29]. Moreover, the white-box knowledge setting of decomposition methods may make them impractical for users who can only query GNNs (e.g., using cloud-based GNN services [168]). Finally, a potential weakness of existing generation methods is that they ignore counterfactual explanations (i.e., they only consider the relationship between explanations and the labels of input graph data in their reward or loss functions).…”

Section: Discussionmentioning

confidence: 99%

“…However, both preprocessing and in-processing methods are practical only when users are permitted to modify graph data and GNN models, respectively. If users can only treat GNNs as black boxes (e.g., when using cloud-based GNN services [168]), then users can only query GNNs and employ post-processing methods (e.g., InFoRM-R [33]) to alleviate predication unfairness in target GNNs.…”

Section: Discussionmentioning

confidence: 99%

“…Of the existing explainers, the methods that require black-box knowledge of GNNs (see Section IV-C) present stronger practicality, as users sometimes cannot obtain enough knowledge about GNN systems to support a white-box approach. For example, when using cloud-based GNN services [168], users can only query GNN services to obtain predictions on the input graph data, and cannot access the inner backward gradients of GNNs, which are indispensable for the training of some explainers (e.g., GNNExplainer [22] and PGExplainer [56]). Due to the diverse application scenarios of post-hoc explainers, designing compelling and low-demand explainers represents a promising avenue for facilitating the explainability of practical GNN systems.…”

Section: Future Directions Of Explainable Gnnsmentioning

confidence: 99%

“…In practice, GNN systems are used as black boxes that provide few transparency to either developers or users. For example, cloud-based GNN services perform their functions in the cloud [168]; their underlying operations cannot be directly accessed by end users. In the absence of proper accountability, users will not fully trust GNN service providers, since they have concerns due to frequent security incidents on the cloud [242].…”

Section: Accountability Of Gnnsmentioning

confidence: 99%

See 3 more Smart Citations

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Zhang¹,

Wu²,

Yuan³

et al. 2022

Preprint

View full text Add to dashboard Cite

Graph neural networks (GNNs) have emerged as a series of competent graph learning methods for diverse real-world scenarios, ranging from daily applications like recommendation systems and question answering to cutting-edge technologies such as drug discovery in life sciences and n-body simulation in astrophysics. However, task performance is not the only requirement for GNNs. Performance-oriented GNNs have exhibited potential adverse effects like vulnerability to adversarial attacks, unexplainable discrimination against disadvantaged groups, or excessive resource consumption in edge computing environments. To avoid these unintentional harms, it is necessary to build competent GNNs characterised by trustworthiness. To this end, we propose a comprehensive roadmap to build trustworthy GNNs from the view of the various computing technologies involved. In this survey, we introduce basic concepts and comprehensively summarise existing efforts for trustworthy GNNs from six aspects, including robustness, explainability, privacy, fairness, accountability, and environmental well-being. Additionally, we highlight the intricate cross-aspect relations between the above six aspects of trustworthy GNNs. Finally, we present a thorough overview of trending directions for facilitating the research and industrialisation of trustworthy GNNs.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Future Directions Of Explainable Gnnsmentioning

confidence: 99%

Section: Accountability Of Gnnsmentioning

confidence: 99%

See 2 more Smart Citations

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Zhang¹,

Wu²,

Yuan³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Edge computing has merits in reducing latency [6], personalizing services [13,14,20], resource optimization [15,18], and strengthening privacy and security [23,29,32]. Edge computing for recommendation is still a nascent research area [3,7,9,36,37]. [9] monitor users' timely multi-intentions on edge, such as whether a particular user will buy some goods within an hour, via binary classification.…”

Section: Related Workmentioning

confidence: 99%

Intelligent Request Strategy Design in Recommender System

Qian¹,

Xu²,

Lv³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Waterfall Recommender System (RS), a popular form of RS in mobile applications, is a stream of recommended items consisting of successive pages that can be browsed by scrolling. In waterfall RS, when a user finishes browsing a page, the edge (e.g., mobile phones) would send a request to the cloud server to get a new page of recommendations, known as the paging request mechanism. RSs typically put a large number of items into one page to reduce excessive resource consumption from numerous paging requests, which, however, would diminish the RSs' ability to timely renew the recommendations according to users' real-time interest and lead to a poor user experience. Intuitively, inserting additional requests inside pages to update the recommendations with a higher frequency can alleviate the problem. However, previous attempts, including only non-adaptive strategies (e.g., insert requests uniformly), would eventually lead to resource overconsumption. To this end, we envision a new learning task of edge intelligence named Intelligent Request Strategy Design (IRSD). It aims to improve the effectiveness of waterfall RSs by determining the appropriate occasions of request insertion based on users' real-time intention. Moreover, we propose a new paradigm of adaptive request insertion strategy named Uplift-based On-edge Smart Request Framework (AdaRequest). AdaRequest 1) captures the dynamic change of users' intentions by matching their real-time behaviors with their historical interests based on attention-based neural networks. 2) estimates the counterfactual uplift of user purchase brought by an inserted request based on causal inference. 3) determines the final request insertion strategy by maximizing the utility function under online resource constraints. We conduct extensive experiments on both offline dataset and online A/B test to verify the effectiveness of AdaRequest. Remarkably, AdaRequest has been deployed on the Waterfall RS of Taobao and brought over 3% lift on Gross Merchandise Value (GMV).

show abstract

Opt-CoInfer: Optimal collaborative inference across IoT and cloud for fast and accurate CNN inference

Zhang

Wang

2023

Journal of King Saud University - Computer and Information Scie

View full text Add to dashboard Cite

Edge-Cloud Polarization and Collaboration: A Comprehensive Survey for AI

Cited by 37 publications

References 236 publications

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Trustworthy Graph Neural Networks: Aspects, Methods and Trends

Intelligent Request Strategy Design in Recommender System

Opt-CoInfer: Optimal collaborative inference across IoT and cloud for fast and accurate CNN inference

Contact Info

Product

Resources

About