Christopher J. Riederer

Linking Users Across Domains with Location Data: Theory and Validation

Linking accounts of the same user across datasets -- even when personally identifying information is removed or unavailable -- is an important open problem studied in many contexts. Beyond many practical applications, (such as cross domain analysis, recommendation, and link prediction), understanding this problem more generally informs us on the privacy implications of data disclosure. Previous work has typically addressed this question using either different portions of the same dataset or observing the same behavior across thematically similar domains. In contrast, the general cross-domain case where users have different profiles independently generated from a common but unknown pattern raises new challenges, including difficulties in validation, and remains under-explored. In this paper, we address the reconciliation problem for location-based datasets and introduce a robust method for this general setting. Location datasets are a particularly fruitful domain to study: such records are frequently produced by users in an increasing number of applications and are highly sensitive, especially when linked to other datasets. Our main contribution is a generic and self-tunable algorithm that leverages any pair of sporadic location-based datasets to determine the most likely matching between the users it contains. While making very general assumptions on the patterns of mobile users, we show that the maximum weight matching we compute is provably correct. Although true cross-domain datasets are a rarity, our experimental evaluation uses two entirely new data collections, including one we crawled, on an unprecedented scale. The method we design outperforms naive rules and prior heuristics. As it combines both sparse and dense properties of location-based data and accounts for probabilistic dynamics of observation, it can be shown to be robust even when data gets sparse.

Video: Procrastinator: pacing mobile apps' usage of the network

Generations of computer programmers are taught to prefetch network objects in computer science classes. In practice, prefetching can be harmful to the users wallet when she is on a limited or pay-per-byte cellular data plan. Many popular, professionally-written smartphone apps today prefetch large amounts of network data that the typical user may never use. We present Procrastinator, which automatically decides when to fetch each network object that an app requests. This decision is made based on whether the user is on Wi-Fi or cellular, how many bytes are remaining on the users data plan, and whether the object is needed at the present time. Procrastinator does not require app developer effort, nor app source code, nor OS changes -- it modifies the app binary to trap specific system calls and inject custom code. Our system can achieve as little as no savings to 4X reduction in total bytes transferred by an app, depending on the user and the app. These savings for the data-poor user come with a 300ms median latency penalty on LTE.

Magnetic input for mobile virtual reality

Modern smartphones can create compelling virtual reality (VR) experiences through the use of VR enclosures, devices that encase the phone and project stereoscopic renderings through lenses into the users eyes. Since the touch screen in such designs is typically hidden inside an enclosure, the main interaction mechanism of the device is not accessible. We present a new magnetic input mechanism for mobile VR devices which is wireless, unpowered, inexpensive, provides physical feedback, requires no calibration, and works reliably on the majority of modern smartphones. This is the main input mechanism for Google Cardboard, of which there are over one million units. We show robust gesture recognition, at an accuracy of greater than 95% across smartphones and assess the capabilities, accuracy and limitations of our technique through a user study.

Give in to procrastination and stop prefetching

Generations of computer programmers are taught to prefetch network objects in computer science classes. In practice, prefetching can be harmful to the users wallet when she is on a limited or pay-per-byte cellular data plan. Many popular, professionally-written smartphone apps today prefetch large amounts of network data that the typical user may never use. We present Procrastinator, which automatically decides when to fetch each network object that an app requests. This decision is made based on whether the user is on Wi-Fi or cellular, how many bytes are remaining on the users data plan, and whether the object is needed at the present time. Procrastinator does not require developer effort, nor app source code, nor OS changes -- it modifies the app binary to trap specific system calls and inject custom code. Our system can achieve as little as no savings to 4X savings in bytes transferred, depending on the user and the app. In theory, we can achieve 17X savings, but we need to overcome additional technical challenges.

I don't have a photograph, but you can have my footprints.: Revealing the Demographics of Location Data

Location data are routinely available to a plethora of mobile apps and third party web services. The resulting datasets are increasingly available to advertisers for targeting and also requested by governmental agencies for law enforcement purposes. While the re-identification risk of such data has been widely reported, the discriminative power of mobility has received much less attention. In this study we fill this void with an open and reproducible method. We explore how the growing number of geotagged footprints left behind by social network users in photosharing services can give rise to inferring demographic information from mobility patterns. Chiefly among those, we provide the first detailed analysis of ethnic mobility patterns in two metropolitan areas. This analysis allows us to examine questions pertaining to spatial segregation and the extent to which ethnicity can be inferred using only location data. Our results reveal that even a few location records at a coarse grain can be sufficient for simple algorithms to draw an accurate inference. Our method generalizes to other features, such as gender, offering for the first time a general approach to evaluate discriminative risks associated with location-enabled personalization.

Algorithmic Glass Ceiling in Social Networks: The effects of social recommendations on network diversity

As social recommendations such as friend suggestions and people to follow become increasingly popular and influential on the growth of social media, we find that prominent social recommendation algorithms can exacerbate the under-representation of certain demographic groups at the top of the social hierarchy. To study this imbalance in online equal opportunities, we leverage new Instagram data and offer for the first time an analysis that studies the effect of gender, homophily and growth dynamics under social recommendations. Our mathematical analysis demonstrates the existence of an algorithmic glass ceiling that exhibits all the properties of the metaphorical social barrier that hinders groups like women or people of color from attaining equal representation. What raises concern is that our proof shows that under fixed minority and homophily parameters the algorithmic effect is systematically larger than the glass ceiling generated by the spontaneous growth of social networks. We discuss ways to address this concern in future design.

Challenges of keyword-based location disclosure

A practical solution to location privacy should be incrementally deployable. We claim it should hence reconcile the economic value of location to aggregators, usually ignored by prior works, with a users control over her information. Location information indeed is being collected and used by many mobile services to improve revenues, and this gives rise to a heated debate: Privacy advocates ask for stricter regulation on information collection, while companies argue that it would jeopardize the thriving economy of the mobile web. We describe a system that gives users control over their information and does not degrade the data given to aggregators. Recognizing that the first challenge is to express locations in a way that is meaningful for advertisers and users, we propose a keyword-based design. Keywords characterize locations, let the users inform the system about their sensitivity to disclosure, and build information directly usable by an advertisers targeting campaign. Our work makes two main contributions: we design a market of location information based on keywords and we analyze its robustness to attacks using data from ad-networks, geo-located services, and cell networks.

Monte Carlo simulation to determine conditions for optical molecular imaging of vascular disease

Atherosclerosis is a significant contributor to cardiovascular disease, which is a leading cause of death in the United States. Optical imaging is emerging as a promising molecular imaging tool to detect the biomarkers of vascular disease and plaque vulnerability. We describe a Monte Carlo method of investigating the use of targeted fluorescent nanoparticles as a reporter agent for optical molecular imaging. In particular, we used the model to provide insight into the optimum excitation and emission wavelengths for in vivo imaging in the presence of blood and auto fluorescence.