A range of applications, from predicting the spread of human and electronic viruses to city planning and resource management in mobile communications, depend on our ability to foresee the whereabouts and mobility of individuals, raising a fundamental question: To what degree is human behavior predictable? Here we explore the limits of predictability in human dynamics by studying the mobility patterns of anonymized mobile phone users. By measuring the entropy of each individual's trajectory, we find a 93% potential predictability in user mobility across the whole user base. Despite the significant differences in the travel patterns, we find a remarkable lack of variability in predictability, which is largely independent of the distance users cover on a regular basis.

http://absimage.aps.org/image/MAR10/MWS_MAR10-2009-007155.pdf

Limits of predictability in human mobility

Contact tracing is an essential public health tool for controlling epidemic disease outbreaks such as the COVID-19 pandemic. Digital contact tracing using real-time locations or proximity of individuals can be used to significantly speed up and scale up contact tracing. In this article, we present our project, REACT, for REAal-time Contact Tracing and risk monitoring via privacy-enhanced tracking of users' locations and symptoms. With privacy enhancement that allows users to control and refine the precision with which their information will be collected and used, REACT will enable: 1) contact tracing of individuals who are exposed to infected cases and identification of hot-spot locations, 2) individual risk monitoring based on the locations they visit and their contact with others; and 3) community risk monitoring and detection of early signals of community spread. We will briefly describe our ongoing work and the approaches we are taking as well as some challenges we encountered in deploying the app.

REACT: real-time contact tracing and risk monitoring using privacy-enhanced mobile tracking

Recent advances in catalytic oxidative desulfurization of fuel oil – A review

We scrutinized the effect of supports (types and calcination temperatures) in Ru/support catalysts in the hydrogenolysis of low-density polyethylene and found that zirconia-supported Ru (Ru/ZrO2) with high-temperature (1073 K) calcined ZrO2 was an effective and reusable catalyst, showing about 3-fold higher activity on catalyst amount basis than the previously reported Ru/CeO2 catalyst with similar selectivites. The catalyst was applicable to the reactions of various polyolefins including high-density polyethylene and polypropylene. Experimental studies and catalyst characterizations such as TPR, XRD, TEM and XAS with various Ru/ZrO2 and reference catalysts such as Ru/CeO2, Ru/SiO2 and Ru/γ-Al2O3 showed good correlations between the Ru particle sizes and catalytic performance in Ru catalysts except for Ru/γ-Al2O3: The catalyst with a moderate size of ~2.5 nm Ru particles provided the highest conversion. The activity per surface Ru metal was higher at larger Ru particle sizes. The selectivity to cheap gas products was low at middle to small Ru particle sizes (1–7 nm), but high at large Ru particles (>7 nm). 

Structure-activity relationship in hydrogenolysis of polyolefins over Ru/support catalysts

A Survey Of differential privacy-based techniques and their applicability to location-Based services

Identifying mobility behaviors in rich trajectory data is of great economic and social interest to various applications including urban planning, marketing and intelligence. Existing work on trajectory clustering often relies on similarity measurements that utilize raw spatial and/or temporal information of trajectories. These measures are incapable of identifying similar moving behaviors that exhibit varying spatiotemporal scales of movement. In addition, the expense of labeling massive trajectory data is a barrier to supervised learning models. To address these challenges, we propose an unsupervised neural approach for mobility behavior clustering, called the Deep Embedded TrajEctory ClusTering network (DETECT). DETECT operates in three parts: first it transforms the trajectories by summarizing their critical parts and augmenting them with context derived from their geographical locality (e.g., using POIs from gazetteers). In the second part, it learns a powerful representation of trajectories in the latent space of behaviors, thus enabling a clustering function (such as k-means) to be applied. Finally, a clustering oriented loss is directly built on the embedded features to jointly perform feature refinement and cluster assignment, thus improving separability between mobility behaviors. Exhaustive quantitative and qualitative experiments on two real-world datasets demonstrate the effectiveness of our approach for mobility behavior analyses.

DETECT: Deep Trajectory Clustering for Mobility-Behavior Analysis

Location-based apps provide users with personalized services tailored to their geographical position. This is highly-beneficial for mobile users, who are able to find points of interest close to their location, or connect with nearby friends. However, sharing location data with service providers also introduces privacy concerns. An adversary with access to fine-grained user locations can infer private details about individuals. Geo-indistinguishability (GeoInd) adapts the popular differential privacy (DP) model to make it suitable for protecting users’ location information. However, existing techniques that implement GeoInd have major drawbacks. Some solutions, such as the planar Laplace mechanism, significantly lower data utility by adding excessive noise. Other approaches, such as the optimal mechanism, achieve good utility, but only work for small sets of candidate locations due to the use of computationally-expensive linear programming. In most cases, locations are used to answer online queries, so a quick response time is essential. In this paper, we propose a technique that achieves GeoInd and scales to large datasets while preserving data utility. Our central idea is to use the composability property of GeoInd to create a multiple-step algorithm that can be used in conjunction with a spatial index. We preserve utility by applying accurate GeoInd mechanisms and we achieve scalability by pruning the solution search space with the help of the index when seeking high-utility outcomes. Our extensive performance evaluation on top of real location datasets from social media apps shows that the proposed technique outperforms significantly the benchmark in terms of utility and/or computational overhead.

A Utility-Preserving and Scalable Technique for Protecting Location Data with Geo-Indistinguishability.

Identifying mobility behaviors in rich trajectory data is of great economic and social interest to various applications including urban planning, marketing and intelligence. Existing work on trajectory clustering often relies on similarity measurements that utilize raw spatial and/or temporal information of trajectories. These measures are incapable of identifying similar moving behaviors that exhibit varying spatio-temporal scales of movement. In addition, the expense of labeling massive trajectory data is a barrier to supervised learning models. To address these challenges, we propose an unsupervised neural approach for mobility behavior clustering, called the Deep Embedded TrajEctory ClusTering network (DETECT). DETECT operates in three parts: first it transforms the trajectories by summarizing their critical parts and augmenting them with context derived from their geographical locality (e.g., using POIs from gazetteers). In the second part, it learns a powerful representation of trajectories in the latent space of behaviors, thus enabling a clustering function (such as $k$-means) to be applied. Finally, a clustering oriented loss is directly built on the embedded features to jointly perform feature refinement and cluster assignment, thus improving separability between mobility behaviors. Exhaustive quantitative and qualitative experiments on two real-world datasets demonstrate the effectiveness of our approach for mobility behavior analyses.

Contact tracing is an essential public health tool for controlling epidemic disease outbreaks such as the COVID-19 pandemic. Digital contact tracing using real-time locations or proximity of individuals can be used to significantly speed up and scale up contact tracing. In this demonstration, we present our system, REACT, for REAl-time Contact Tracing and risk monitoring via privacy-enhanced tracking of users’ locations. With privacy enhancement that allows users to control and refine the precision with which their information will be collected and used, REACT will enable: 1) contact tracing of individuals who are exposed to infected cases and identification of hot-spot locations, 2) individual risk monitoring based on the locations they visit and their contact with others. In this paper, we demonstrate the procedure of contact tracing using our application and the utility of contact tracing given the protected locations.

Ritesh Ahuja

Papers

REACT: real-time contact tracing and risk monitoring using privacy-enhanced mobile tracking

DETECT: Deep Trajectory Clustering for Mobility-Behavior Analysis

A Utility-Preserving and Scalable Technique for Protecting Location Data with Geo-Indistinguishability.

DETECT: Deep Trajectory Clustering for Mobility-Behavior Analysis

REACT: Real-Time Contact Tracing and Risk Monitoring via Privacy-Enhanced Mobile Tracking