发布求助
文献互助 智能选刊 最新文献

Proceedings of the Workshop on Distributed Infrastructures for Deep Learning最新文献

筛选
英文 中文
Tools and Techniques for Privacy-aware, Edge-centric Distributed Deep Learning 隐私感知、边缘中心分布式深度学习的工具和技术
Proceedings of the Workshop on Distributed Infrastructures for Deep Learning Pub Date : 2020-12-07 DOI: 10.1145/3429882.3430105
Ziran Min, Robert Canady, Uttam Ghosh, A. Gokhale, Akram Hakiri
{"title":"Tools and Techniques for Privacy-aware, Edge-centric Distributed Deep Learning","authors":"Ziran Min, Robert Canady, Uttam Ghosh, A. Gokhale, Akram Hakiri","doi":"10.1145/3429882.3430105","DOIUrl":"https://doi.org/10.1145/3429882.3430105","url":null,"abstract":"Training and inferencing phases of Deep Learning (DL) are compute-intensive that require substantial amount of cloud-hosted resources. However, real-time needs of some edge-based applications as well as the variable and wildly fluctuating edge-cloud latency require new ways to exploit clusters of edge devices in a decentralized and federated manner that perform on-device or edge-based DL training/inferencing. However, edge-based DL is fraught with many challenges including the need to discover the right resources, heterogeneity in resource types leading to non-uniform execution times among cluster members, increased incidences of failures and network disconnectivity leading to consistency issues, preserving privacy of data used in DL tasks, the type of distributed DL algorithm used and its performance on the chosen resources, and many others. To address this plethora of challenges, this paper proposes solutions, which include privacy-preserving, inferencing on heterogeneous edge devices, and CAP-driven trade-offs and Blockchain-based consensus.","PeriodicalId":424477,"journal":{"name":"Proceedings of the Workshop on Distributed Infrastructures for Deep Learning","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130559832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
Graph Representation Matters in Device Placement 图形表示在设备放置中很重要
Proceedings of the Workshop on Distributed Infrastructures for Deep Learning Pub Date : 2020-12-07 DOI: 10.1145/3429882.3430104
Milko Mitropolitsky, Zainab Abbas, A. H. Payberah
{"title":"Graph Representation Matters in Device Placement","authors":"Milko Mitropolitsky, Zainab Abbas, A. H. Payberah","doi":"10.1145/3429882.3430104","DOIUrl":"https://doi.org/10.1145/3429882.3430104","url":null,"abstract":"Modern Neural Network (NN) models require more data and parameters to perform ever more complicated tasks. One approach to train a massive NN is to distribute it across multiple devices. This approach raises a problem known as the device placement problem. Most of the state-of-the-art solutions that tackle this problem leverage graph embedding techniques. In this work, we assess the impact of different graph embedding techniques on the quality of device placement, measured by (i) the execution time of partitioned NN models, and (ii) the computation time of the graph embedding technique. In particular, we expand Placeto, a state-of-the-art device placement solution, and evaluate the impact of two graph embedding techniques, GraphSAGE and P-GNN, compared to the original Placeto graph embedding model, Placeto-GNN. In terms of the execution time improvement, we achieve an increase of 23.967% when using P-GNN compared to Placeto-GNN, while GraphSAGE produces 1.165% better results than Placeto-GNN. Regarding computation time, GraphSAGE has a gain of 11.569% compared to Placeto-GNN, whereas P-GNN is 6.95% slower than it.","PeriodicalId":424477,"journal":{"name":"Proceedings of the Workshop on Distributed Infrastructures for Deep Learning","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130383282","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
Challenges and Opportunities of DNN Model Execution Caching DNN模型执行缓存的挑战与机遇
Proceedings of the Workshop on Distributed Infrastructures for Deep Learning Pub Date : 2019-12-09 DOI: 10.1145/3366622.3368147
Guin Gilman, Samuel S. Ogden, R. Walls, Tian Guo
{"title":"Challenges and Opportunities of DNN Model Execution Caching","authors":"Guin Gilman, Samuel S. Ogden, R. Walls, Tian Guo","doi":"10.1145/3366622.3368147","DOIUrl":"https://doi.org/10.1145/3366622.3368147","url":null,"abstract":"We explore the opportunities and challenges of model execution caching, a nascent research area that promises to improve the performance of cloud-based deep inference serving. Broadly, model execution caching relies on servers that are geographically close to the end-device to service inference requests, resembling a traditional content delivery network (CDN). However, unlike a CDN, such schemes cache execution rather than static objects. We identify the key challenges inherent to this problem domain and describe the similarities and differences with existing caching techniques. We further introduce several emergent concepts unique to this domain, such as memory-adaptive models and multi-model hosting, which allow us to make dynamic adjustments to the memory requirements of model execution.","PeriodicalId":424477,"journal":{"name":"Proceedings of the Workshop on Distributed Infrastructures for Deep Learning","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115542632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 5
A sim2real framework enabling decentralized agents to execute MADDPG tasks 一个sim2real框架，允许分散代理执行MADDPG任务
Proceedings of the Workshop on Distributed Infrastructures for Deep Learning Pub Date : 2019-12-09 DOI: 10.1145/3366622.3368146
Young-Ho Suh, Sungpil Woo, Hyunhak Kim, Dong-Hwan Park
{"title":"A sim2real framework enabling decentralized agents to execute MADDPG tasks","authors":"Young-Ho Suh, Sungpil Woo, Hyunhak Kim, Dong-Hwan Park","doi":"10.1145/3366622.3368146","DOIUrl":"https://doi.org/10.1145/3366622.3368146","url":null,"abstract":"Multi-agent RL is a process of training the agents to collaborate with others. We argue that an additional 'reality gap' in the system aspects occurs when applying sim2real to the multi-agent RL, especially when performing the 'transferred' collaborative task in the real-world environment. In this paper, we propose an ADO framework enabling decentralized agents to participate in performing collaborative tasks without suffering from the reality gap. Our contribution is threefold. First, we clearly identify and summarize the reality gaps in the context of the sim2real of multi-agent RL. Second, we propose a new system model to deal with system issues derived from when executing collaborative tasks. Third, we design and implement a software framework to support system issues required in developing and executing collaborative tasks in the real world.","PeriodicalId":424477,"journal":{"name":"Proceedings of the Workshop on Distributed Infrastructures for Deep Learning","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125572489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 3
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整，为了账户安全，请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
请完成安全验证×
相关产品
×
本文献相关产品
联系我们：info@booksci.cn Book学术提供免费学术资源搜索服务，方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号：604180095
Book学术官方微信
小红书