arXiv subjects

Kristina Lerman

Publications and source records attributed to Kristina Lerman.

At least 127 records · Page 7Linked to original sources

arXiv2020

Having a Bad Day? Detecting the Impact of Atypical Life Events Using Wearable Sensors

Life events can dramatically affect our psychological state and work performance. Stress, for example, has been linked to professional dissatisfaction, increased anxiety, and workplace burnout. We explore the impact of positive and negative life events on a number of psychological constructs through a multi-month longitudinal study of hospital and aerospace workers. Through causal inference, we demonstrate that positive life events increase positive affect, while negative events increase stress, anxiety and negative affect. While most events have a transient effect on psychological states, major negative events, like illness or attending a funeral, can reduce positive affect for multiple days. Next, we assess whether these events can be detected through wearable sensors, which can cheaply and unobtrusively monitor health-related factors. We show that these sensors paired with embedding-based learning models can be used ``in the wild'' to capture atypical life events in hundreds of workers across both datasets. Overall our results suggest that automated interventions based on physiological sensing may be feasible to help workers regulate the negative effects of life events.

cs.HC↗

arXiv2020

Learning Behavioral Representations from Wearable Sensors

Continuous collection of physiological data from wearable sensors enables temporal characterization of individual behaviors. Understanding the relation between an individual's behavioral patterns and psychological states can help identify strategies to improve quality of life. One challenge in analyzing physiological data is extracting the underlying behavioral states from the temporal sensor signals and interpreting them. Here, we use a non-parametric Bayesian approach to model sensor data from multiple people and discover the dynamic behaviors they share. We apply this method to data collected from sensors worn by a population of hospital workers and show that the learned states can cluster participants into meaningful groups and better predict their cognitive and psychological states. This method offers a way to learn interpretable compact behavioral representations from multivariate sensor signals.

cs.LG↗

arXiv2020

Tracking Social Media Discourse About the COVID-19 Pandemic: Development of a Public Coronavirus Twitter Data Set

At the time of this writing, the novel coronavirus (COVID-19) pandemic outbreak has already put tremendous strain on many countries' citizens, resources and economies around the world. Social distancing measures, travel bans, self-quarantines, and business closures are changing the very fabric of societies worldwide. With people forced out of public spaces, much conversation about these phenomena now occurs online, e.g., on social media platforms like Twitter. In this paper, we describe a multilingual coronavirus (COVID-19) Twitter dataset that we have been continuously collecting since January 22, 2020. We are making our dataset available to the research community (https://github.com/echen102/COVID-19-TweetIDs). It is our hope that our contribution will enable the study of online conversation dynamics in the context of a planetary-scale epidemic outbreak of unprecedented proportions and implications. This dataset could also help track scientific coronavirus misinformation and unverified rumors, or enable the understanding of fear and panic -- and undoubtedly more. Ultimately, this dataset may contribute towards enabling informed solutions and prescribing targeted policy interventions to fight this global crisis.

cs.SI↗

arXiv2020

Unequal Impact and Spatial Aggregation Distort COVID-19 Growth Rates

The COVID-19 pandemic has emerged as a global public health crisis. To make decisions about mitigation strategies and to understand the disease dynamics, policy makers and epidemiologists must know how the disease is spreading in their communities. We analyze confirmed infections and deaths over multiple geographic scales to show that COVID-19's impact is highly unequal: many subregions have nearly zero infections, and others are hot spots. We attribute the effect to a Reed-Hughes-like mechanism in which disease arrives at different times and grows exponentially. Hot spots, however, appear to grow faster than neighboring subregions and dominate spatially aggregated statistics, thereby amplifying growth rates. The staggered spread of COVID-19 can also make aggregated growth rates appear higher even when subregions grow at the same rate. Public policy, economic analysis and epidemic modeling need to account for potential distortions introduced by spatial aggregation.

q-bio.QM↗

arXiv2020

Challenges in Forecasting Malicious Events from Incomplete Data

The ability to accurately predict cyber-attacks would enable organizations to mitigate their growing threat and avert the financial losses and disruptions they cause. But how predictable are cyber-attacks? Researchers have attempted to combine external data -- ranging from vulnerability disclosures to discussions on Twitter and the darkweb -- with machine learning algorithms to learn indicators of impending cyber-attacks. However, successful cyber-attacks represent a tiny fraction of all attempted attacks: the vast majority are stopped, or filtered by the security appliances deployed at the target. As we show in this paper, the process of filtering reduces the predictability of cyber-attacks. The small number of attacks that do penetrate the target's defenses follow a different generative process compared to the whole data which is much harder to learn for predictive models. This could be caused by the fact that the resulting time series also depends on the filtering process in addition to all the different factors that the original time series depended on. We empirically quantify the loss of predictability due to filtering using real-world data from two organizations. Our work identifies the limits to forecasting cyber-attacks from highly filtered data.

cs.CR↗

arXiv2020

Predictability limit of partially observed systems

Applications from finance to epidemiology and cyber-security require accurate forecasts of dynamic phenomena, which are often only partially observed. We demonstrate that a system's predictability degrades as a function of temporal sampling, regardless of the adopted forecasting model. We quantify the loss of predictability due to sampling, and show that it cannot be recovered by using external signals. We validate the generality of our theoretical findings in real-world partially observed systems representing infectious disease outbreaks, online discussions, and software development projects. On a variety of prediction tasks---forecasting new infections, the popularity of topics in online discussions, or interest in cryptocurrency projects---predictability irrecoverably decays as a function of sampling, unveiling fundamental predictability limits in partially observed systems.

cs.CY↗

arXiv2019

Learning Fair and Interpretable Representations via Linear Orthogonalization

To reduce human error and prejudice, many high-stakes decisions have been turned over to machine algorithms. However, recent research suggests that this does not remove discrimination, and can perpetuate harmful stereotypes. While algorithms have been developed to improve fairness, they typically face at least one of three shortcomings: they are not interpretable, their prediction quality deteriorates quickly compared to unbiased equivalents, and they are not easily transferable across models. To address these shortcomings, we propose a geometric method that removes correlations between data and any number of protected variables. Further, we can control the strength of debiasing through an adjustable parameter to address the trade-off between prediction quality and fairness. The resulting features are interpretable and can be used with many popular models, such as linear regression, random forest, and multilayer perceptrons. The resulting predictions are found to be more accurate and fair compared to several state-of-the-art fair AI algorithms across a variety of benchmark datasets. Our work shows that debiasing data is a simple and effective solution toward improving fairness.

cs.CY↗

arXiv2019

Collaboration Drives Individual Productivity

How does the number of collaborators affect individual productivity? Results of prior research have been conflicting, with some studies reporting an increase in individual productivity as the number of collaborators grows, while other studies showing that the {free-rider effect} skews the effort invested by individuals, making larger groups less productive. The difference between these schools of thought is substantial: if a super-scaling effect exists, as suggested by former studies, then as groups grow, their productivity will increase even faster than their size, super-linearly improving their efficiency. We address this question by studying two planetary-scale collaborative systems: GitHub and Wikipedia. By analyzing the activity of over 2 million users on these platforms, we discover that the interplay between group size and productivity exhibits complex, previously-unobserved dynamics: the productivity of smaller groups scales super-linearly with group size, but saturates at larger sizes. This effect is not an artifact of the heterogeneity of productivity: the relation between group size and productivity holds at the individual level. People tend to do more when collaborating with more people. We propose a generative model of individual productivity that captures the non-linearity in collaboration effort. The proposed model is able to explain and predict group work dynamics in GitHub and Wikipedia by capturing their maximally informative behavioral features, and it paves the way for a principled, data-driven science of collaboration.

cs.HC↗

arXiv2019

The Transsortative Structure of Networks

Network topologies can be non-trivial, due to the complex underlying behaviors that form them. While past research has shown that some processes on networks may be characterized by low-order statistics describing nodes and their neighbors, such as degree assortativity, these quantities fail to capture important sources of variation in network structure. We introduce a property called transsortativity that describes correlations among a node's neighbors, generalizing these statistics from immediate one-hop neighbors to two-hop neighbors. We describe how transsortativity can be systematically varied, independently of the network's degree distribution and assortativity. Moreover, we show that it can significantly impact the spread of contagions as well as the perceptions of neighbors, known as the majority illusion. Our work improves our ability to create and analyze more realistic models of complex networks.

physics.soc-ph↗

arXiv2019

Quantifying the Impact of Cognitive Biases in Question-Answering Systems

Crowdsourcing can identify high-quality solutions to problems; however, individual decisions are constrained by cognitive biases. We investigate some of these biases in an experimental model of a question-answering system. In both natural and controlled experiments, we observe a strong position bias in favor of answers appearing earlier in a list of choices. This effect is enhanced by three cognitive factors: the attention an answer receives, its perceived popularity, and cognitive load, measured by the number of choices a user has to process. While separately weak, these effects synergistically amplify position bias and decouple user choices of best answers from their intrinsic quality. We end our paper by discussing the novel ways we can apply these findings to substantially improve how high-quality answers are found in question-answering systems.

cs.HC↗

arXiv2019

Massive Multi-Agent Data-Driven Simulations of the GitHub Ecosystem

Simulating and predicting planetary-scale techno-social systems poses heavy computational and modeling challenges. The DARPA SocialSim program set the challenge to model the evolution of GitHub, a large collaborative software-development ecosystem, using massive multi-agent simulations. We describe our best performing models and our agent-based simulation framework, which we are currently extending to allow simulating other planetary-scale techno-social systems. The challenge problem measured participant's ability, given 30 months of meta-data on user activity on GitHub, to predict the next months' activity as measured by a broad range of metrics applied to ground truth, using agent-based simulation. The challenge required scaling to a simulation of roughly 3 million agents producing a combined 30 million actions, acting on 6 million repositories with commodity hardware. It was also important to use the data optimally to predict the agent's next moves. We describe the agent framework and the data analysis employed by one of the winning teams in the challenge. Six different agent models were tested based on a variety of machine learning and statistical methods. While no single method proved the most accurate on every metric, the broadly most successful sampled from a stationary probability distribution of actions and repositories for each agent. Two reasons for the success of these agents were their use of a distinct characterization of each agent, and that GitHub users change their behavior relatively slowly.

cs.MA↗

arXiv2019

MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing

Existing popular methods for semi-supervised learning with Graph Neural Networks (such as the Graph Convolutional Network) provably cannot learn a general class of neighborhood mixing relationships. To address this weakness, we propose a new model, MixHop, that can learn these relationships, including difference operators, by repeatedly mixing feature representations of neighbors at various distances. Mixhop requires no additional memory or computational complexity, and outperforms on challenging baselines. In addition, we propose sparsity regularization that allows us to visualize how the network prioritizes neighborhood information across different graph datasets. Our analysis of the learned architectures reveals that neighborhood mixing varies per datasets.

cs.LG↗

arXiv2019

Discovering Hidden Structure in High Dimensional Human Behavioral Data via Tensor Factorization

In recent years, the rapid growth in technology has increased the opportunity for longitudinal human behavioral studies. Rich multimodal data, from wearables like Fitbit, online social networks, mobile phones etc. can be collected in natural environments. Uncovering the underlying low-dimensional structure of noisy multi-way data in an unsupervised setting is a challenging problem. Tensor factorization has been successful in extracting the interconnected low-dimensional descriptions of multi-way data. In this paper, we apply non-negative tensor factorization on a real-word wearable sensor data, StudentLife, to find latent temporal factors and group of similar individuals. Meta data is available for the semester schedule, as well as the individuals' performance and personality. We demonstrate that non-negative tensor factorization can successfully discover clusters of individuals who exhibit higher academic performance, as well as those who frequently engage in leisure activities. The recovered latent temporal patterns associated with these groups are validated against ground truth data to demonstrate the accuracy of our framework.

cs.LG↗

arXiv2019

Friendship Paradox Biases Perceptions in Directed Networks

How popular a topic or an opinion appears to be in a network can be very different from its actual popularity. For example, in an online network of a social media platform, the number of people who mention a topic in their posts---i.e., its global popularity---can be dramatically different from how people see it in their social feeds---i.e., its perceived popularity---where the feeds aggregate their friends' posts. We trace the origin of this discrepancy to the friendship paradox in directed networks, which states that people are less popular than their friends (or followers) are, on average. We identify conditions on network structure that give rise to this perception bias, and validate the findings empirically using data from Twitter. Within messages posted by Twitter users in our sample, we identify topics that appear more frequently within the users' social feeds, than they do globally, i.e., among all posts. In addition, we present a polling algorithm that leverages the friendship paradox to obtain a statistically efficient estimate of a topic's global prevalence from biased perceptions of individuals. We characterize the bias of the polling estimate, provide an upper bound for its variance, and validate the algorithm's efficiency through synthetic polling experiments on our Twitter data. Our paper elucidates the non-intuitive ways in which the structure of directed networks can distort social perceptions and resulting behaviors.

cs.SI↗

arXiv2019

Characterizing Activity on the Deep and Dark Web

The deep and darkweb (d2web) refers to limited access web sites that require registration, authentication, or more complex encryption protocols to access them. These web sites serve as hubs for a variety of illicit activities: to trade drugs, stolen user credentials, hacking tools, and to coordinate attacks and manipulation campaigns. Despite its importance to cyber crime, the d2web has not been systematically investigated. In this paper, we study a large corpus of messages posted to 80 d2web forums over a period of more than a year. We identify topics of discussion using LDA and use a non-parametric HMM to model the evolution of topics across forums. Then, we examine the dynamic patterns of discussion and identify forums with similar patterns. We show that our approach surfaces hidden similarities across different forums and can help identify anomalous events in this rich, heterogeneous data.

cs.CY↗

arXiv2018

Characterizing the 2016 Russian IRA Influence Campaign

Until recently, social media were seen to promote democratic discourse on social and political issues. However, this powerful communication ecosystem has come under scrutiny for allowing hostile actors to exploit online discussions in an attempt to manipulate public opinion. A case in point is the ongoing U.S. Congress investigation of Russian interference in the 2016 U.S. election campaign, with Russia accused of, among other things, using trolls (malicious accounts created for the purpose of manipulation) and bots (automated accounts) to spread propaganda and politically biased information. In this study, we explore the effects of this manipulation campaign, taking a closer look at users who re-shared the posts produced on Twitter by the Russian troll accounts publicly disclosed by U.S. Congress investigation. We collected a dataset of 13 million election-related posts shared on Twitter in the year of 2016 by over a million distinct users. This dataset includes accounts associated with the identified Russian trolls as well as users sharing posts in the same time period on a variety of topics around the 2016 elections. We use label propagation to infer the users' ideology based on the news sources they share. We are able to classify a large number of users as liberal or conservative with precision and recall above 84%. Conservative users who retweet Russian trolls produced significantly more tweets than liberal ones, about 8 times as many in terms of tweets. Additionally, trolls' position in the retweet network is stable over time, unlike users who retweet them who form the core of the election-related retweet network by the end of 2016. Using state-of-the-art bot detection techniques, we estimate that about 5% and 11% of liberal and conservative users are bots, respectively.

cs.SI↗

arXiv2018

Contagions in Social Networks: Effects of Monophilic Contagion, Friendship Paradox and Reactive Networks

We consider SIS contagion processes over networks where, a classical assumption is that individuals' decisions to adopt a contagion are based on their immediate neighbors. However, recent literature shows that some attributes are more correlated between two-hop neighbors, a concept referred to as monophily. This motivates us to explore monophilic contagion, the case where a contagion (e.g. a product, disease) is adopted by considering two-hop neighbors instead of immediate neighbors (e.g. you ask your friend about the new iPhone and she recommends you the opinion of one of her friends). We show that the phenomenon called friendship paradox makes it easier for the monophilic contagion to spread widely. We also consider the case where the underlying network stochastically evolves in response to the state of the contagion (e.g. depending on the severity of a flu virus, people restrict their interactions with others to avoid getting infected) and show that the dynamics of such a process can be approximated by a differential equation whose trajectory satisfies an algebraic constraint restricting it to a manifold. Our results shed light on how graph theoretic consequences affect contagions and, provide simple deterministic models to approximate the collective dynamics of contagions over stochastic graph processes.

cs.SI↗

arXiv2018

Predicting and Explaining Behavioral Data with Structured Feature Space Decomposition

Modeling human behavioral data is challenging due to its scale, sparseness (few observations per individual), heterogeneity (differently behaving individuals), and class imbalance (few observations of the outcome of interest). An additional challenge is learning an interpretable model that not only accurately predicts outcomes, but also identifies important factors associated with a given behavior. To address these challenges, we describe a statistical approach to modeling behavioral data called the structured sum-of-squares decomposition (S3D). The algorithm, which is inspired by decision trees, selects important features that collectively explain the variation of the outcome, quantifies correlations between the features, and partitions the subspace of important features into smaller, more homogeneous blocks that correspond to similarly-behaving subgroups within the population. This partitioned subspace allows us to predict and analyze the behavior of the outcome variable both statistically and visually, giving a medium to examine the effect of various features and to create explainable predictions. We apply S3D to learn models of online activity from large-scale data collected from diverse sites, such as Stack Exchange, Khan Academy, Twitter, Duolingo, and Digg. We show that S3D creates parsimonious models that can predict outcomes in the held-out data at levels comparable to state-of-the-art approaches, but in addition, produces interpretable models that provide insights into behaviors. This is important for informing strategies aimed at changing behavior, designing social systems, but also for explaining predictions, a critical step towards minimizing algorithmic bias.

cs.CY↗