報告地點：西北工業大學長安校區計算機學院105會議室
報告時間：2019年5月6日上午10：30-11：30
報告人：香港中文大學John C.S. Lui教授, IEEE/ACM Fellow
邀請人：褚偉波副教授
報告題目：一類基于在線學習的網絡多路徑選擇方法
報告人簡介:
呂自成教授目前是香港中文大學計算機科學與工程系的李卓敏榮譽教授。他的研究方向聚焦于機器學習在網絡科學、網絡經濟學、網絡/系統安全、大規模分布式系統和性能測評理論方面的研究與應用。呂教授獲得了諸多教學與科研方面的獎項，包括香港中文大學校長模范教學獎和香港中文大學職員杰出研究獎（2011-2012）。他獲得了IFIP WG 7.3 Performance 2005, IEEE/IFIP NOMS 2006，SIMPLEX'14，ACM RecSys’17等重要國際會議的最佳學術論文獎以及ACM Mobihoc’18和ASONAM’17會議的最佳論文提名獎。他是IFIP WG 7.3，ACM，IEEE等多個重要協會的會士，Croucher基金會的高級研究專家，以及現任的ACM SIGMETRICS會議主席。呂教授的個人興趣包括電影和閱讀。
報告摘要:
過去十年間，接入計算機網絡的主機數量呈現出爆炸式增長。多種用于主機間數據傳輸的多路徑協議被相繼提出。然而，當前的數據傳輸多路徑協議由于忽略了網絡傳輸延遲、可用帶寬和數據丟包等隨機特性而使它們在應用上收到了極大的限制。另外，許多應用對網絡傳輸延遲、鏈路帶寬和丟包率等有一定要求。本報告將介紹一種基于網絡特性在線學習的多路徑選取框架，用于滿足不同應用對數據傳輸的需求。具體地，將介紹一類用于分別滿足最大傳輸延遲、網絡帶寬和丟包率約束的在線學習多路徑選取算法，并在理論上確保算法具有次線性的regret和violation兩個關鍵性能指標。
Biography:
John C.S. Lui is currently the Choh-Ming Li Professor of the Computer Science & Engineering Department at The Chinese University of Hong Kong. His current research interests are in machine learning on network sciences, network economics, network/system security (e.g., cloud security, mobile security, ...etc), large scale distributed systems and performance evaluation theory. John received various departmental teaching awards and the CUHK Vice-Chancellor's Exemplary Teaching Award, as well as the CUHK Faculty of Engineering Research Excellence Award (2011-2012). He is a co-recipient of the IFIP WG 7.3 Performance 2005, IEEE/IFIP NOMS 2006 and SIMPLEX'14 Best Paper Awards, ACM RecSys’17 best paper award and best paper runner-up in ACM Mobihoc’18 and ASONAM’17. He is an elected member of the IFIP WG 7.3, Fellow of ACM, Fellow of IEEE, Senior Research Fellow of the Croucher Foundation and is currently the chair of the ACM SIGMETRICS. His personal interests include films and general reading.
Title:
An Online Learning Multi-path Selection Framework for Multi-path Transmission Protocols
Abstract:
In the last decade, we have witnessed a tremendous growth of inter-connectivity among hosts in networks. Many new data transmission protocols have been developed to enable multi-path data transmissions between two hosts. However, the existing multi-path transmission protocol designs are limited as they neglect the stochastic nature of the metrics of the paths, e.g., latency, available bandwidth, and packet loss. Moreover, there are different design requirements in the applications, such as low latency, bandwidth throttling, and low loss rate in data delivery. In this talk, we propose a flexible online learning multi-path selection (OLMPS) framework to select multiple paths by learning the stochastic metrics of the paths and meeting the design requirements of the applications. Specifically, we design a set of novel online learning algorithms in the OLMPS framework for three different applications -- maxRTT constrained, bandwidth constrained, and loss rate constrained, multi-path selection, to select paths and satisfy the requirements. We prove that the algorithms can provide theoretical guarantees on both sublinear regret and sublinear violation in our OLMPS framework.
