Sun E.,Cornell Technology
CSCW 2017 - Companion of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing | Year: 2017
Through my doctoral research, I aim to gain a deeper understanding of how LBPHDs, location-based, post-hoc data applications, can be used to build social capital in urban communities. From a case study of a dating app that utilizes shared location history, happn, we explored how LBPHD information has been used to build interpersonal relationships. Based on these findings, we designed Move-Meant, an application that extends LBPHD from interpersonal to community-level information sharing. Preliminary qualitative field results suggest the potential of MoveMeant to increase local community awareness through dissemination of local knowledge and discovery of third places. Future work in this area include a larger qualitative study and quantitative study of MoveMeant, and further understanding its application to other situations.
Sun E.,Cornell Technology |
De Oliveira R.,YouTube |
Proceedings of the ACM Conference on Computer Supported Cooperative Work, CSCW | Year: 2017
In order to better understand social aspects of the short-fonn video watching experience, we investigated the journey to cowatching, from searching and discovering content, to choosing and experiencing videos with others. After identifying, through a large-scale survey, some of the most typical situations that bring people to YouTube, we deployed a one weeklong diary study with 12 participants in which they performed a set of frequent video tasks at their leisure, half by themselves, and half with someone else. Following the diary study, we had participants reenact the diary study tasks remotely with the experimenter. We observed that users face multiple challenges on the journey to co-watching a video. They must share a device designed for an individual, use different methods for selecting videos than when by themselves, negotiate or tum-take in order to make a decision, and potentially watch a video that they do not enjoy. Along this journey, users must engage in impression management to consider how their choices might make them appear to others. We present design recommendations for remote and collocated co-watching to improve the social video watching experience. © 2017 ACM.
McLachlan R.,Cornell Technology |
Opila C.,Cornell Technology |
Shah N.,Cornell Technology |
Sun E.,Cornell Technology |
Naaman M.,Cornell Technology
Conference on Human Factors in Computing Systems - Proceedings | Year: 2016
Peer to peer sharing of physical goods in local communities seems like a promising concept, but platforms that offer these services have not yet reached critical mass. Based on a preliminary analysis of 15 interviews with residents of NYC about sharing in the local community, our results suggest that there is a disconnect between the kind of items that people would like to borrow and those that people would be willing to share. While people are most interested in expensive, infrequently used items, they indicated concern over liability for damages and trust of strangers. We discuss the trade-offs of introducing insurance through the platforms themselves and suggest potential alternate ways of facilitating exchanges of these items. © 2016 Authors.
Bindschaedler V.,University of Illinois at Urbana - Champaign |
Shokri R.,Cornell Technology
Proceedings - 2016 IEEE Symposium on Security and Privacy, SP 2016 | Year: 2016
Camouflaging user's actual location with fakes is a prevalent obfuscation technique for protecting location privacy. We show that the protection mechanisms based on the existing (ad hoc) techniques for generating fake locations are easily broken by inference attacks. They are also detrimental to many utility functions, as they fail to credibly imitate the mobility of living people. This paper introduces a systematic approach to synthesizing plausible location traces. We propose metrics that capture both geographic and semantic features of real location traces. Based on these statistical metrics, we design a privacy-preserving generative model to synthesize location traces which are plausible to be trajectories of some individuals with consistent lifestyles and meaningful mobilities. Using a state-of-the-art quantitative framework, we show that our synthetic traces can significantly paralyze location inference attacks. We also show that these fake traces have many useful statistical features in common with real traces, thus can be used in many geo-data analysis tasks. We guarantee that the process of generating synthetic traces itself is privacy preserving and ensures plausible deniability. Thus, although the crafted traces statistically resemble human mobility, they do not leak significant information about any particular individual whose data is used in the synthesis process. © 2016 IEEE.
Merolla P.A.,IBM |
Arthur J.V.,IBM |
Alvarez-Icaza R.,IBM |
Cassidy A.S.,IBM |
And 16 more authors.
Science | Year: 2014
Inspired by the brain's structure, we have developed an efficient, scalable, and flexible non-von Neumann architecture that leverages contemporary silicon technology. To demonstrate, we built a 5.4-billion-transistor chip with 4096 neurosynaptic cores interconnected via an intrachip network that integrates 1 million programmable spiking neurons and 256 million configurable synapses. Chips can be tiled in two dimensions via an interchip communication interface, seamlessly scaling the architecture to a cortexlike sheet of arbitrary size. The architecture is well suited to many applications that use complex neural networks in real time, for example, multiobject detection and classification. With 400-pixel-by-240-pixel video input at 30 frames per second, the chip consumes 63 milliwatts.
Lingel J.,Microsoft |
Naaman M.,Cornell Technology |
Proceedings of the ACM Conference on Computer Supported Cooperative Work, CSCW | Year: 2014
We use qualitative interviews with 26 transnational migrants in New York City to analyze socio-technical practices related to online identity work. We focus specifically on the use of Facebook, where benefits included keeping in touch with friends and family abroad and documenting everyday urban life. At the same time, many participants also reported experiences of fatigue, socio-cultural tensions and concerns about maintaining a sense of personal privacy. These experiences highlight how transnational practices complicate context collapse, where the geographic dispersal of participants' personal networks renders visible conflicts of "flattened" online networks. Our findings also suggest a kind of technology-enabled codeswitching, where transnational migrants leverage social media to perform identities that alternate between communities, nationalities and geographies. This analysis informs HCI research on transnationalism and technological practices, and helps expose the complexities of online identity work in terms of shifting social and spatial contexts. Copyright © 2014 ACM.
Shokri R.,University of Texas at Austin |
Shmatikov V.,Cornell Technology
Proceedings of the ACM Conference on Computer and Communications Security | Year: 2015
Deep learning based on artificial neural networks is a very popular approach to modeling, classifying, and recognizing complex data such as images, speech, and text. The unprecedented accuracy of deep learning methods has turned them into the foundation of new AI-based services on the Internet. Commercial companies that collect user data on a large scale have been the main beneficiaries of this trend since the success of deep learning techniques is directly proportional to the amount of data available for training. Massive data collection required for deep learning presents obvious privacy issues. Users' personal, highly sensitive data such as photos and voice recordings is kept indefinitely by the companies that collect it. Users can neither delete it, nor restrict the purposes for which it is used. Furthermore, centrally kept data is subject to legal subpoenas and extra-judicial surveillance. Many data owners-for example, medical institutions that may want to apply deep learning methods to clinical records-are prevented by privacy and confidentiality concerns from sharing the data and thus benefitting from large-scale deep learning. In this paper, we design, implement, and evaluate a practical system that enables multiple parties to jointly learn an accurate neuralnetwork model for a given objective without sharing their input datasets. We exploit the fact that the optimization algorithms used in modern deep learning, namely, those based on stochastic gradient descent, can be parallelized and executed asynchronously. Our system lets participants train independently on their own datasets and selectively share small subsets of their models' key parameters during training. This offers an attractive point in the utility/privacy tradeoff space: participants preserve the privacy of their respective data while still benefitting from other participants' models and thus boosting their learning accuracy beyond what is achievable solely on their own inputs. We demonstrate the accuracy of our privacypreserving deep learning on benchmark datasets.
Juels A.,Cornell Technology
Proceedings of ACM Symposium on Access Control Models and Technologies, SACMAT | Year: 2014
Decoy objects, often labeled in computer security with the term honey, are a powerful tool for compromise detection and mitigation. There has been little exploration of overarching theories or set of principles or properties, however. This short paper (and accompanying keynote talk) briefly explore two properties of honey systems, indistinguishability and secrecy. The aim is to illuminate a broad design space that might encompass a wide array of areas in information security, including access control, the main topic of this symposium.
Pass R.,Cornell Technology
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2016
Cryptographic notions of knowledge consider the knowledge obtained, or possessed, by computationally-bounded agents under adversarial conditions. In this talk, we will survey some recent cryptographically-inspired approaches for reasoning about agents in the context of game-theory and mechanism design (where agents typically are modelled as computationally unbounded). © Springer International Publishing Switzerland 2016.
Shmatikov V.,Cornell Technology
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2015
SSL/TLS is the de facto standard for secure Internet communications. Deployed widely in Web browsers and non-browser software, it is intended to provide end-to-end security even against active, man-in-the-middle attacks. This security fundamentally depends on correct validation of X.509 certificates presented when the connection is established. I will first demonstrate that many SSL/TLS deployments are completely insecure against man-in-the-middle attacks. Vulnerable software includes cloud computing clients, merchant SDKs responsible for transmitting payment information from e-commerce sites to payment processors, online shopping software, and many forms of middleware. Even worse, several popular SSL/TLS implementations do not validate certificates correctly and thus all software based on them is generically insecure. These bugs affect even common Web browsers, where minor validation errors such as recent certificate expiration can mask serious issues such as failure to authenticate the Web server’s identity. I will then analyze the root causes of these vulnerabilities and describe how we used “frankencerts,” a new methodology for automatically testing SSL/TLS implementations, to uncover dozens of subtle certificate validation bugs in popular SSL/TLS implementations. © Springer International Publishing Switzerland 2015.