April 2016 - talks - lists.cs.princeton.edu

Sergiy Popovych will be presenting his General Exam on Tuesday, May 3, 2016 at 2pm in CS 402.
by Nicki Gotsis 29 Apr '16

29 Apr '16

Sergiy Popovych will be presenting his General Exam on Tuesday, May 3, 2016 at 2pm in CS 402. The members of his committee are David August (adviser), Nick Feamster and Arvind Narayanan. Everyone is invited to attend his talk, and those faculty wishing to remain for the oral exam following are welcome to do so. His abstract and reading list follow below. As has been demonstrated time and again, almost all systems can be compromised. Connecting to an external network through a unidirectional gateway physically disallows any outgoing traffic, and thus prevents data theft. Existing unidirectional gateways act as stateless links, only supporting inherently unidirectional applications. In an attempt to support applications that require request/response behavior (e.g. web browsing), Starlight adds a second, bidirectionally connected machine not entrusted with sensitive information, which improves usability but makes the system vulnerable to attacks involving the user (e.g. phishing, insider attacks). This work demonstrates the benefit of making unidirectional gateways stateful. The proposed solution, Moonshine, uses this state to pre-cache content from the external network, after which an illusion of request/reply behavior can be achieved, thus partially supporting such applications as web browsing and software updates without compromising security. Books: [1] Ross J. Anderson. 2008. Security Engineering: A Guide to Building Dependable Distributed Systems (2 ed.). Wiley Publishing. (Chapters 2, 3, 4, 6, 8, 9, 21) [2] Asaf Shabtai, Yuval Elovici, Lior Rokach. 2012. A Survey of Data Leakage Detection and Prevention Solutions. Springer US. Papers: [1] J. H. Saltzer, D. P. Reed, and D. D. Clark. 1984. End-to-end arguments in system design. ACM Trans. Comput. Syst. 2, 4 (November 1984), 277-288. [2] David L. Chaum. 1981. Untraceable electronic mail, return addresses, and digital pseudonyms. Commun. ACM 24, 2 (February 1981), 84-90. [3] M. H. Kang and I. S. Moskowitz, "A Pump for rapid, reliable, secure communication", Proc. ACM Conf. Computer Commun. Security, pp. 119-129, 1993. [4] M. Anderson, C. North, J. Griffin, R. Milner, J. Yesberg and K. Yiu, "Starlight: Interactive Link," Computer Security Applications Conference, 1996., 12th Annual, San Diego, CA, 1996, pp. 55-63. [5] R. T. Barker and C. J. Cheese, "The application of data diodes for securely connecting nuclear power plant safety systems to the corporate IT network," System Safety, incorporating the Cyber Security Conference 2012, 7th IET International Conference on, Edinburgh, 2012, pp. 1-6. [6] B. S. Jeon and J. C. Na, "A study of cyber security policy in industrial control system using data diodes," 2016 18th International Conference on Advanced Communication Technology (ICACT), Pyeongchang, 2016, pp. 314-317. [7] J. Epstein, "Fifteen Years after TX: A Look Back at High Assurance Multi-Level Secure Windowing," 2006 22nd Annual Computer Security Applications Conference (ACSAC'06), Miami Beach, FL, 2006, pp. 301-320. [8] Eric Byres. 2013. The air gap: SCADA's enduring security myth. Commun. ACM 56, 8 (August 2013), 29-31 [9] Butler W. Lampson. 1973. A note on the confinement problem. Commun. ACM 16, 10 (October 1973), 613-615. [10] A. Wool. 2004. A Quantitative Study of Firewall Configuration Errors. Computer 37, 6 (June 2004), 62-67.

1 1

Xiaozhou Li will present his FPO on Thursday, 5/5/2016 at 2pm in CS 302 "Systems and Algorithms for High-Performance, Cost-Efficient Key-Value Storage"
by Nicki Gotsis 28 Apr '16

28 Apr '16

Xiaozhou Li will present his FPO on Thursday, 5/5/2016 at 2pm in CS 302 "Systems and Algorithms for High-Performance, Cost-Efficient Key-Value Storage" The members of his committee are Michael Freedman (adviser); Kyle Jamieson and Michael Kaminsky (Intel) (Readers); Jennifer Rexford and Kai Li (examiners). A copy of his thesis is available in Room 310. Everyone is invited to attend his talk. The talk abstract follow below. Key-value storage systems are increasingly essential building blocks of modern cloud and big data applications. The workloads these systems support often require random access to small objects over massive datasets with highly skewed and dynamic key popularity. It is challenging for a storage cluster to serve these workloads with both high performance and low-cost operations. Today’s systems usually sacrifice one for the other. In this dissertation, we present novel approaches to improve both the performance and cost-efficiency of key-value systems by combining new hardware and software techniques with careful architectural design and algorithmic optimizations. First, at cluster scale, we build SwitchKV, a heterogeneous system that uses small highend cache nodes to guarantee load balancing across many SSD-based backend nodes under nearly-arbitrary workloads. The cache nodes absorb the hottest queries so that no individual backend node is over-burdened or underutilized. The system exploits OpenFlow switches to enable efficient content-aware routing so that it can achieve scalable high throughput, low tail latency, and high availability. It uses new algorithms to keep the cache and switch forwarding rules updated with low overhead, and to ensure stable high performance under rapidly changing workloads. SwitchKV can meet the service level objectives for many cloud services more efficiently than traditional systems. Second, to improve the efficiency of each individual multi-core server, we build a highthroughput and memory-efficient concurrent hash table based around optimistic cuckoo hashing. Our re-design minimizes critical section length, reduces interprocessor coherence traffic, and enables effective prefetching through careful algorithm and data structure engineering. We explore hardware transactional memory and fine-grained locking for concurrency control, and find that both of them require the same level of algorithmic efforts to achieve high performance. Our new hash table design greatly outperforms other optimized concurrent hash tables for both read- and write-heavy workloads, even while using substantially less memory for small key-value items.

1 0

Swati Roy will present her General Exam Wednesday, May 4, 2016 in CS 302 at 2:30pm
by Nicki Gotsis 27 Apr '16

27 Apr '16

The members of her committee are Nick Feamster (Adviser), Jennifer Rexford, and Kyle Jamieson. Everyone is invited to attend her talk, and those faculty wishing to remain for the oral exam following are welcome to do so. Please find her abstract and reading list below. Abstract: As smartphone users increasingly rely on cellular networks to access voice, video, and web applications, guaranteeing good performance and high availability is more important than ever. Historically, managing cellular network configuration has been a manual, error-prone process; recently, however, automated solutions such as SON (Self Organizing Networks) controllers are being deployed for dynamic tuning of network configuration. Self-organizing networks (SON) are a new paradigm in cellular networks that dynamically react to changes in network and traffic conditions, automatically adjusting the network configuration to improve performance for end users. For example, when a cell tower becomes congested, a SON controller might automatically tune the transmission power or antenna tilt of neighboring cell towers to better distribute load. Large cellular network providers are currently deploying SON in their cellular networks to improve end-user performance for voice, video, and web applications. SON automates many aspects of cellular network configuration, but it is nonetheless susceptible to software bugs. Additionally, if it is configured incorrectly, it may provide suboptimal performance for particular network conditions and traffic load. One of the biggest challenges in SON configuration is determining how a configuration change affects the performance that end users experience. Existing systems are based on classic time-series prediction, and compare the performance between a study group (where a particular configuration change was implemented) and a control group (where the change was not implemented) to tackle the impact of external factors or rely on comparisons of a particular configuration change against a baseline. Yet, because a SON controller operates in a dynamic environment, isolating the effects of any single configuration change against a control group or a baseline is incredibly difficult. We propose a capability (Veracity) to analyze and quantify the performance effects of SON actions. Assessing the effects of SON control is difficult because of the dynamic nature of SON and the dependency of end-user performance on factors such as radio channel quality, mobility and traffic load. Veracity addresses these using model-driven impact detection and quantification. Our evaluation using data collected from an operational cellular network demonstrates that Veracity is accurate. Veracity is now being used by the service providers' field operation teams for the assessment of SON effectiveness in arenas and stadiums. Reading List: [1] Ajay Mahimkar, Han Hee Song, Zihui Ge, Aman Shaikh, Jia Wang, Jennifer Yates, Yin Zhang, Joanne Emmons: Detecting the performance impact of upgrades in large operational networks. In ACM SIGCOMM, 2010. [2] A. Mahimkar, Z. Ge, J. Yates, C. Hristov, V. Cordaro, S. Smith, J. Xu, and M. Stockert: Robust assessment of changes in cellular networks. In ACM CoNEXT, 2013. [3] M. Shafiq, L. Ji, A. Liu, J. Pang, S. Venkataraman, and J. Wang: A first look at cellular network performance during crowded events. In SIGMETRICS, 2013. [4] M. Zubair Shafiq, Lusheng Ji, Alex X. Liu, Jia Wang: Characterizing and Modeling Internet Traffic Dynamics of Cellular Devices, In SIGMETRICS, 2011. [5] S. Sundaresan, W. de Donato, N. Feamster, R. Teixeira, S. Crawford, and A. Pescape: Broadband Internet Performance: A View From the Gateway . In SIGCOMM, Aug. 2011. [6] Srikanth Sundaresan , Nick Feamster , Renata Teixeira: Measuring the Performance of User Traffic in Home Wireless Networks. In PAM, 2015 [7] Mingzhe Li, Feng Li, Mark Claypool, and Robert Kinicki: Weather Forecasting - Predicting Performance for Streaming Video over Wireless LANs, In NOSSDAV, June 2005. [8] Naga Katta, Haoyu Zhang, Michael Freedman, and Jennifer Rexford. Ravana: Controller Fault-Tolerance in Software-Defined Networking, In SIGCOMM Symposium For SDN Research (SOSR), June 2015. [9] Athula Balachandran, Vaneet Aggarwal, Emir Halepovic, Jeffrey Pang, Srinivasan Seshan, Shobha Venkataraman, He Yan: Modeling Web Quality-of-Experience on Cellular Networks. In MobiCom 2014. [10] Xin Jin, Li Erran Li, Laurent Vanbever, and Jennifer Rexford. SoftCell: Scalable and Flexible Cellular Core Network Architecture. In CoNext 2013. [11] L. L. Peterson and B. S. Davie. Computer Networks, Fifth Edition: A Systems Approach. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 5th edition, 2011. Best, Swati

1 1

Yixin Sun will present her General Exam on Monday, May 2, 2016 at 10:15am in CS 402
by Nicki Gotsis 25 Apr '16

25 Apr '16

Yixin Sun will present her General Exam on Monday, May 2, 2016 at 10:15am in CS 402. The members of her committee are Mung Chiang & Prateek Mittal (advisers), Nick Feamster, and Arvind Narayanan. Everyone is invited to attend her talk, and those faculty wishing to remain for the oral exam following are welcome to do so. Her abstract and reading list follow below. Abstract: The Tor network is a widely used system for anonymous communication. However, the vulnerability of the Tor network to adversaries who can monitor Internet traffic, such as Autonomous Systems (ASes), is not well understood. We present a suite of new attacks, called Raptor, that can be launched by Autonomous Systems (ASes) to compromise user anonymity. First, AS-level adversaries can exploit the asymmetric nature of Internet routing to increase the chance of observing at least one direction of user traffic at both ends of the communication. Second, strategic adversaries can manipulate Internet routing via BGP hijacks (to discover the users using specific Tor guard nodes) and interceptions (to perform traffic analysis). We demonstrate the feasibility of Raptor attacks by performing real-world attacks on the live Tor network, while ensuring that we do not harm real users. Third, we measure the vulnerability of the Tor network to such active BGP attacks. We show that more than 90% of BGP prefixes hosting Tor relays are vulnerable to more-specific prefix attacks, and some Autonomous Systems that carry a large portion of Tor traffic, such as OVH, also have low resilience to equally-specific prefix attacks. Finally, we present the BGP monitoring system that detects routing anomalies on the Tor network in real time. We evaluate the system on live BGP data and show that it has no false positives while successfully detects an attack injected by us. Reading list: Computer Networks (5th Edition) by TANENBAUM and WETHERALL. Chapters 1, 5-8. Security Engineering (2nd Edition) by Anderson. Chapters 1-8, 21-24. 1. Ballani, H., Francis, P. and Zhang, X., 2007, August. A study of prefix hijacking and interception in the Internet. In ACM SIGCOMM Computer Communication Review (Vol. 37, No. 4, pp. 265-276). ACM. 2. Lad, M., Oliveira, R., Zhang, B. and Zhang, L., 2007, June. Understanding resiliency of internet topology against prefix hijack attacks. In Dependable Systems and Networks, 2007. DSN'07. 37th Annual IEEE/IFIP International Conference on (pp. 368-377). IEEE. 3. Lad, M., Massey, D., Pei, D., Wu, Y., Zhang, B. and Zhang, L., 2006, August. PHAS: A Prefix Hijack Alert System. In Usenix Security. 4. Dingledine, R., Mathewson, N. and Syverson, P., 2004. Tor: The second-generation onion router. Naval Research Lab Washington DC. 5. Johnson, A., Wacek, C., Jansen, R., Sherr, M. and Syverson, P., 2013, November. Users get routed: Traffic correlation on Tor by realistic adversaries. In Proceedings of the 2013 ACM SIGSAC conference on Computer & communications security (pp. 337-348). ACM. 6. Murdoch, S.J. and Zieliński, P., 2007, June. Sampled traffic analysis by internet-exchange-level adversaries. In Privacy Enhancing Technologies (pp. 167-183). Springer Berlin Heidelberg. 7. Akhoondi, M., Yu, C. and Madhyastha, H.V., 2012, May. LASTor: A low-latency AS-aware Tor client. In Security and Privacy (SP), 2012 IEEE Symposium on (pp. 476-490). IEEE. 8. Edman, M. and Syverson, P., 2009, November. AS-awareness in Tor path selection. In Proceedings of the 16th ACM conference on Computer and communications security (pp. 380-389). ACM. 9. Feamster, N. and Dingledine, R., 2004, October. Location diversity in anonymity networks. In Proceedings of the 2004 ACM workshop on Privacy in the electronic society (pp. 66-76). ACM. 10. Starov, O., Nithyanand, R., Zair, A., Gill, P. and Schapira, M., 2015. Measuring and mitigating AS-level adversaries against Tor. arXiv preprint arXiv:1505.05173.

1 0

Fwd: EE Seminar - April 28, 2016, 11:00 am, E-Quad B205 - Dr. Zhichun Li
by Jennifer Rexford 25 Apr '16

25 Apr '16

> > From: Sania Sadhvani <sadhvani(a)Princeton.EDU> > Subject: EE Seminar - April 28, 2016, 11:00 am, E-Quad B205 - Dr. Zhichun Li > Date: April 25, 2016 at 2:54:21 PM EDT > To: ee-seminar(a)Princeton.EDU > Reply-To: Sania Sadhvani <sadhvani(a)Princeton.EDU> > > > DEPARTMENT OF > ELECTRICAL ENGINEERING SEMINAR SERIES > > > > > Speaker: Zhichun Li > NEC Research Labs > Title: Counter Advanced Persistent Threat through Big-Data Enabled Security > Intelligence > Date: April 28, 2016 > Time: 11:00 a.m. > Room: B205 Engineering Quadrangle > Host: Prof. Prateek Mittal > > Abstract: Today serious security incidents keep appearing in mainstream media. Cyber-attack is no longer just for fun but has grown up and become a large dark industry. Data breaches from Target, Home Depot, JP Morgan Chase, Sony, eBay, OPM.gov etc., not only cause severe impact to the business or government agency, but also affect millions of people. The attacks behind the scene are so call Advanced Persistent Threats (APTs). In this talk, I will introduce what is an APT, and what characteristics those attacks have by going through a case study on how Target got hacked. Then, I will examine why APT is hard to detect by existing industry security solutions, and potentially how we can do better. To fight against APT, we started the Automated Security Intelligence (ASI) project in 2012, and later found DARPA Transparent Computing (TC) program BAA in 2014 shared the same vision, which is “connecting the dots” across multiple activities that are individually not suspicious enough, but collectively indicate malice or abnormal behavior. The ASI project achieves ubiquitous monitoring inside enterprises with Windows, Linux and Mac agents, includes a big-data middleware to handle the massive events collected, and provides a platform that supports various security applications. In particular, in this talk, I will highlight research progress from two early efforts on “connecting the dots”: temporal behavior query language, which enables a search tool for security analysts to quickly explore the potential linkage among events, and activity backtracking, which is a semi-automatic tool on answering “how this happens”. > > Biography: Zhichun (ZL) Li is a senior researcher at NEC Research Labs in Princeton, NJ. At NEC Labs, he manages the newly formed Computer Security Department. He has initiated and led the large-scale interdisciplinary research project called Automated Security Intelligence involving 15 researchers from Princeton and Tokyo. He has broad research interests in the areas of security and systems with an emphasis on enterprise security with big-data, smartphone security, network security etc. Before joining NEC Labs, he received his Ph.D. on Dec 2009 from Northwestern University. He earned both M.S. and B.S. degrees from Tsinghua University in China. Previously, he has conducted research at Microsoft Research Redmond and International Computer Science Institute (ICSI) in UC Berkeley. > > > > > > > > > > >

1 0

Colloquium Speaker Ben Shneiderman, Today, 12:30pm
by Nicole E. Wagenblast 22 Apr '16

22 Apr '16

Colloquium Speaker Ben Shneiderman, University of Maryland Friday, April 22, 12:30pm Computer Science 105 The New ABCs of Research: Achieving Breakthrough Collaborations All Princeton ACM / IEEE-CS meetings are open to the public. Students and their parents are welcome. How should we organize research programs in computer science and technology? Research helps us produce innovative new products (the next iPhone), and research also delivers “foundational theories” (new algorithms and supporting technologies). Solving the immense problems of the 21st century will require ambitious research teams that are skilled at producing practical solutions and foundational theories simultaneously – that is the ABC Principle: Applied & Basic Combined. Then these research teams can deliver high-impact outcomes by applying the SED Principle: Blend Science, Engineering and Design Thinking, which encourages use of the methods from all three disciplines. These guiding principles (ABC & SED) are meant to replace Vannevar Bush’s flawed linear model from 1945 that has misled researchers for 70+ years. These new guiding principles will enable students, faculty, business leaders, and government policy makers to accelerate discovery and innovation. The examples in the talk will emphasize how these guiding principles can be applied to reinvigorate computing research. Ben Shneiderman is a Distinguished University Professor in the Department of Computer Science at the University of Maryland. He is also the Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://www.cs.umd.edu/hcil/), and a Member of the University of Maryland Institute for Advanced Computer Studies (UMIACS). He is a Fellow of the AAAS, ACM, IEEE, and NAI, and a Member of the National Academy of Engineering, in recognition of his pioneering contributions to human-computer interaction and information visualization. His contributions include the direct manipulation concept, clickable highlighted web-links, touchscreen keyboards, dynamic query sliders for Spotfire, development of treemaps, novel network visualizations for NodeXL, and temporal event sequence analysis for electronic health records. Ben is the co-author with Catherine Plaisant of Designing the User Interface: Strategies for Effective Human-Computer Interaction (6th ed., 2016). His book Leonardo’s Laptop (MIT Press) won the IEEE book award for Distinguished Literary Contribution. Shneiderman’s latest book is The New ABCs of Research: Achieving Breakthrough Collaborations (Oxford, February 2016).

1 0

Colloquium Speaker Ben Shneiderman, Fri April 22, 12:30pm-reminder
by Nicole E. Wagenblast 21 Apr '16

21 Apr '16

Colloquium Speaker Ben Shneiderman, University of Maryland Friday, April 22, 12:30pm Computer Science 105 The New ABCs of Research: Achieving Breakthrough Collaborations All Princeton ACM / IEEE-CS meetings are open to the public. Students and their parents are welcome. How should we organize research programs in computer science and technology? Research helps us produce innovative new products (the next iPhone), and research also delivers “foundational theories” (new algorithms and supporting technologies). Solving the immense problems of the 21st century will require ambitious research teams that are skilled at producing practical solutions and foundational theories simultaneously – that is the ABC Principle: Applied & Basic Combined. Then these research teams can deliver high-impact outcomes by applying the SED Principle: Blend Science, Engineering and Design Thinking, which encourages use of the methods from all three disciplines. These guiding principles (ABC & SED) are meant to replace Vannevar Bush’s flawed linear model from 1945 that has misled researchers for 70+ years. These new guiding principles will enable students, faculty, business leaders, and government policy makers to accelerate discovery and innovation. The examples in the talk will emphasize how these guiding principles can be applied to reinvigorate computing research. Ben Shneiderman is a Distinguished University Professor in the Department of Computer Science at the University of Maryland. He is also the Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://www.cs.umd.edu/hcil/), and a Member of the University of Maryland Institute for Advanced Computer Studies (UMIACS). He is a Fellow of the AAAS, ACM, IEEE, and NAI, and a Member of the National Academy of Engineering, in recognition of his pioneering contributions to human-computer interaction and information visualization. His contributions include the direct manipulation concept, clickable highlighted web-links, touchscreen keyboards, dynamic query sliders for Spotfire, development of treemaps, novel network visualizations for NodeXL, and temporal event sequence analysis for electronic health records. Ben is the co-author with Catherine Plaisant of Designing the User Interface: Strategies for Effective Human-Computer Interaction (6th ed., 2016). His book Leonardo’s Laptop (MIT Press) won the IEEE book award for Distinguished Literary Contribution. Shneiderman’s latest book is The New ABCs of Research: Achieving Breakthrough Collaborations (Oxford, February 2016).

1 0

Colloquium Speaker: Joseph Bonneau, Today- 12:30pm
by Nicole E. Wagenblast 21 Apr '16

21 Apr '16

Colloquium Speaker Joseph Bonneau , from Stanford University Thursday, April 21- 12:30pm Computer Science 105 Improving security requires both empirically-grounded insights into existing systems and threats, as well as theoretically-grounded solutions that anticipate how future users and attackers will adapt. I will present examples of both. I’ll begin by introducing empirical methods that I created to bring quantitative rigor to the question of how users choose authentication secrets (PINs, passwords, and security questions), a topic that has long been misunderstood due to a lack of data. I'll then present two theoretically-grounded approaches that apply cryptography to providing transparency that trusted authorities are behaving correctly. The first addresses servers for distributing public keys for secure communication, ensuring that the authority cannot lie without being detected. The second ensures that banks that store bitcoins are solvent: that they actually are holding as many bitcoins as they have promised to their clients. Joseph Bonneau is a Postdoctoral Researcher at Stanford University and a Technology Fellow at the Electronic Frontier Foundation. His research focuses on cryptography and security protocols, particularly how they interact with human and organizational behavior and economic incentives. Recently he has focused on Bitcoin and related cryptocurrencies and secure messaging tools. He is also known for his work on passwords and web authentication. He received a PhD from the University of Cambridge under the supervision of Ross Anderson and an BS/MS from Stanford under the supervision of Dan Boneh. Last year he was as a Postdoctoral Fellow at CITP, Princeton and he has previously worked at Google, Yahoo, and Cryptography Research Inc.

1 0

Robert Kiefer will present his FPO, "New Abstractions for Mobile Connectivity and Resource Management" Wednesday, 4/27/2016 at 2:00pm in CS 302.
by Nicki Gotsis 20 Apr '16

20 Apr '16

Robert Kiefer will present his FPO, "New Abstractions for Mobile Connectivity and Resource Management" Wednesday, 4/27/2016 at 2:00pm in CS 302. The members of his committee are Michael Freedman (Adviser); Nick Feamster and Jennifer Rexford (Readers); Margaret Martonosi, David August (Examiners). A copy of his/her thesis, is available in Room 310. Everyone is invited to attend his/her talk. The talk title and abstract follow below. "New Abstractions for Mobile Connectivity and Resource Management" Mobile devices have become an important—and for many, primary—means of connecting to the Internet. They o↵er great flexibility in how and where they are used, even with constraints like battery life and data caps. However, their network stack and operating system are lacking when it comes to network mobility and managing their limited resources. First, mobility, if even addressed by devices and apps, is done in an ad-hoc manner to “hide” connectivity issues rather than fix underlying shortcomings. Second, resource management is dealt with in broad strokes, scattered among various app and device settings pages. In this thesis, we provide new abstractions addressing these issues. To better handle mobility in the network stack, we need to remove the overloading of IP/port as both app-level connection identifiers and network addresses, thereby allowing apps to maintain a connection even when addresses change. To that end, we present ECCP, a protocol that splits the current transport layer into two: one for connection management (e.g., startup, teardown, migration) and one for data delivery semantics (e.g., reliable, best-e↵ort, etc). ECCP is part of a larger network architecture, Serval, which addresses several pain points with today’s networked systems, consisting of replicated backend services and mobile, multi-homed clients. We derive a state machine for ECCP supporting migration and multipath, and through Serval, we demonstrate ECCP’s value in scenarios utilizing mobility (e.g., VM migration) while achieving par performance with TCP. To utilize better mobility support, devices need a more expressive resource management system. Currently, network choice and resource management is limited to broad decisions (e.g., if cellular network usage is allowed). Users must either choose “good enough” settings or else micromanage their device and apps, e.g., turning WiFi on/o↵ while streaming music between hotspots. Instead, we introduce Tango, a programmatic, policy-based network resource management platform, that works at the device and app levels to o↵er more flexiblity. Users and apps define policies that dynamically monitor device state to adjust resource usage on their behalf. Using Tango, a user can, e.g., minimize cellular data while streaming music (using it only to prevent playback pauses), saving up to 60-95%.

1 0

Pixl Talk: Alexandre Alahi, Wednesday Apr 20 1:30PM-2:30PM
by Nicole E. Wagenblast 20 Apr '16

20 Apr '16

Speaker : Alexandre Alahi Time : Wednesday Apr 20 1:30PM-2:30PM Location : CS Room 402 Title : Towards Socially - aware Artificial Intelligence : Human Behavior Understanding at Scale Abstract : Over the past sixty years, Intelligent Machines (IM) have made great progress in playing games, tagging images in isolation, and recently making decisions for self-driving vehicles. Despite these advancements, they are still far from making decisions in social scenes and assisting humans in public spaces such as terminals, malls, campuses, or any crowded urban environment. To overcome these limitations, we need to empower machines with social intelligence , i.e., the ability to get along well with others and facilitate mutual cooperation. This is crucial to design smart spaces that adapt to the behavior of humans for efficiency, or develop autonomous machines that assist in crowded public spaces (e.g., delivery robots, or self-navigating segways). In this talk, I will present my work towards socially - aware machines that can understand human social dynamics and learn to forecast them. First, I will highlight the machine vision techniques behind understanding the behavior of more than 100 million individuals captured by multi-modal cameras in urban spaces. I will show how to use sparsity promoting priors to extract meaningful information about human behavior from an overwhelming volume of high dimensional and high entropy data. Second, I will introduce a new deep learning method to forecast human social behavior . The causality behind human behavior is an interplay between both observable and non-observable cues (e.g., intentions). For instance, when humans walk into crowded urban environments such as a busy train terminal, they obey a large number of (unwritten) common sense rules and comply with social conventions. They typically avoid crossing groups and keep a personal distance to their surrounding. I will present detailed insights on how to learn these interactions from millions of trajectories. I will describe a new recurrent neural network that can jointly reason on correlated sequences and simulate human trajectories in crowded scenes. It opens new avenues of research in learning the causalities behind the world we observe. I will conclude my talk by mentioning some ongoing work in applying these techniques to social robots, and the first generation of smart hospitals. Bio : Alexandre Alahi is currently a postdoctoral fellow at Stanford University and received his PhD from EPFL. His research interests span visual information processing, computer vision, machine learning, robotics, and are focused around understanding and forecasting human social behaviors. In particular, he is interested in sparse approximation, deep learning, big visual data processing, real-time machine vision, and multi-modal reasoning. He was awarded the Swiss NSF early and advanced researcher grants. He won the CVPR 2012 Open Source Award for his work on Retina-inspired image descriptor, and the ICDSC 2009 Challenge Prize for his sparsity driven algorithm to track sport players. His PhD was nominated for the EPFL PhD prize. His research has been covered by the Wall Street journal, and PBS TV channel in the US, as well as European newspapers, Swiss TV news, and Euronews TV channel in Europe. He has also co-founded the startup Visiosafe, transferring his research on understanding human behavior into an industrial product. He was selected as the Top 20 Swiss Venture leaders in 2010 and won several startup competitions.

1 0