[Adapt] FW: 【面试报告会通知】11月18日（周三）9-长聘教轨助理教授岗位面试报告会

Tue Nov 17 11:13:01 CST 2020

You guys are welcome!

Kenny

From: "chen-zx at cs.sjtu.edu.cn" <chen-zx at cs.sjtu.edu.cn>
Date: Tuesday, November 17, 2020 at 11:04 AM
To: "all at cs.sjtu.edu.cn" <all at cs.sjtu.edu.cn>
Subject: 【面试报告会通知】11月18日（周三）9:00-12:00--李卓钊、张牧涵、陈国兴3位博士--长聘教轨助理教授岗位面试报告会

各位老师好！

11月18日（周三）上午9:00-12:00将在3-412会议室召开李卓钊、张牧涵、陈国兴 三位博士的长聘教轨助理教授岗位面试报告会，欢迎感兴趣的老师到场旁听报告，也欢迎转发给学生，可在问答环节中进行学术交流。谢谢！

报告安排：

  9:00-10:00   李卓钊博士学术报告+问答环节

10:00-11:00   张牧涵博士学术报告+问答环节

11:00-12:00   陈国兴博士学术报告+问答环节

--李卓钊博士报告信息--

Title: Efficient and Scalable Systems for End-to-end Computing

Abstract: The emergence of high-speed networks and the exponential growth in internet-connected and embedded devices is pioneering a new era of ubiquitous computing. As we near the end of Moore's Law, while simultaneously dealing with phenomenal growth of data volumes and velocities, there is a rapidly growing need to make use of parallel and distributed systems, from the edge devices near data sources,  and on premises clusters, through to specialized accelerators, supercomputers, and public clouds. Fortunately, high-speed networks and virtualization platforms make it possible for applications to be executed anywhere, where data is generated, cycles are freely available, or computation is cheap. However, due to the heterogeneity (in terms of hardware and interfaces), highly scalable, and geographically distributed nature of modern applications, it is increasingly challenging for developers to develop general solutions, provision infrastructures, and optimize their applications.

To address these challenges we need to rethink programming paradigms and develop new systems that simplify development, offload the burden of infrastructure management, and enable portability across heterogeneous systems such that developers can focus on the problems unique to their applications. Toward these goals I will present recent work developing systems to support parallel programming in Python and distributed function execution across heterogeneous compute resources, and new learning systems to support publication and execution of machine learning models. Finally, I will talk about a PageRank based method to place the demands on execution sites for high utilization and high performance.

Bio: Zhuozhao Li is a Postdoctoral Scholar in the Department of Computer Science at the University of Chicago. His research focuses on networked and distributed computer systems, with an emphasis on building efficient, robust, and scalable systems for computing and developing novel solutions to challenging system management problems. He earned a Ph.D. in the Department of Computer Science at the University of Virginia, 2018. He received a B.S. degree from Zhejiang University and a M.S. degree from the University of Southern California. He was awarded Outstanding Graduate Research Assistant at the University of Virginia and Service Award of IEEE MASS conference. His research was in the Best Paper Finalist at ACM HPDC.

---张牧涵博士报告信息---

Title：Graph Deep Learning: Methods and Applications

Abstract：The past few years have seen the growing prevalence of deep neural networks on various

application domains including computer vision, speech recognition, machine translation, self-driving cars, game playing, bioinformatics, and healthcare etc. In the meanwhile, graph learning has been another hot field among the machine learning and data mining communities, which aims to learn knowledge from graph-structured data. Examples of graph learning range from social network analysis such as community detection and link prediction, to relational machine learning such as knowledge graph completion and recommender systems, to mutli-graph tasks such as graph classification and graph generation etc. 

An emerging new field, graph deep learning, aims at applying deep learning to graphs. To deal with graph-structured data, graph neural networks (GNNs) are invented in recent years which directly take graphs as input and output graph/node representations. In this talk, I will dive deep into graph neural networks and introduce our contributions in this new field. By developing new algorithms, architectures and theories, we push graph neural networks’ boundary to a wide range of graph learning problems, including: 1) graph classification; 2) healthcare representation learning; 3) link prediction; 4) recommender systems; 5) graph generation; and 6) graph structure optimization.

Bio：Dr. Muhan Zhang is currently a research scientist at Facebook AI, working on large-scale graph modeling problems and techniques at Facebook. His research interests lie in graph-related machine learning and data mining problems, especially graph classification, link prediction, recommender systems, healthcare representation learning, and neural architecture search. He received a PhD degree in computer science from Washington University in St. Louis (2015-2019), and a BE degree in electronics engineering from Shanghai Jiao Tong University (2011-2015). Dr. Zhang serves as program committee members for KDD, NeurIPS, ICML, AAAI, IJCAI regularly, and serve as reviewers for TPAMI, TKDE, AOAS, JAIR, TNSE, WWWJ, PlosOne.

---陈国兴博士报告信息---

Title: Side-Channel Attacks and Defenses for Intel SGX

Abstract: Intel Software Guard Extensions (SGX) is an emerging hardware feature that provides software applications a Trusted Execution Environment (TEE) to protect their code and data from untrusted system software. However, prior studies have shown that SGX is vulnerable to side-channel attacks and demonstrated the extraction of secret data, such as sensitive user inputs and cryptographic keys.

This talk will cover some of our works on SGX side-channel attacks and defenses. Particularly, we will first present SgxPectre Attacks, the SGX-variants of the Spectre attacks. SgxPectre Attacks leverage two types of code patterns in the enclave binary. These vulnerable code patterns are found in popular SGX SDKs (e.g., Intel SGX SDK, RustSGX, and Graphene-SGX), indicating any enclaves developed using these SDKs are vulnerable to SgxPectre Attacks. This work was one of the first to demonstrate that Intel SGX’s security guarantees can be completely broken, thus leading to rethinking the security limitations of Intel SGX and other similar TEEs. Second, we will present HyperRace, which is an LLVM-based tool, to close side channels facilitated by Intel Hyper-Threading Technology, such as L1 and L2 caches, branch prediction units (BPU), store buffers and floating-point units (FPU). The idea to create a shadow enclave thread and request the OS to schedule it on the sibling logical core, leaving no room for malicious code to share the same physical core with the logical core running the enclave code. Since the OS is not trusted, we proposed a novel approach for the verification of its scheduling arrangement, by introducing contrived data races. HyperRace is a turn-key solution to the open research problem of defending against Hyper-Threading-enabled side-channel attacks on Intel SGX. Finally, this talk will also outline some of our future research plans and discuss opportunities for collaboration.

Bio: Dr. Guoxing Chen is a Research Scientist at Facebook, Inc. He received a Ph.D. in computer science and engineering from The Ohio State University in 2019. His research interests are in the areas of computer security and privacy, including system security, side channels, and data privacy. He received his B.S. and M.S. degrees from Shanghai Jiao Tong University in 2010 and 2013 respectively.

---

祝好，

陈哲轩

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20201117/9f7bdb12/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 3439 bytes
Desc: not available
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20201117/9f7bdb12/attachment-0003.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.jpg
Type: image/jpeg
Size: 6581 bytes
Desc: not available
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20201117/9f7bdb12/attachment-0004.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image003.jpg
Type: image/jpeg
Size: 5356 bytes
Desc: not available
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20201117/9f7bdb12/attachment-0005.jpg>