Spring 2017 Schedule

4/7 – Dr. Duru Turkoglu – DePaul University
Batch-Responsive Kinetic Data Structures

Kinetic Data Structures (KDSs) provide a framework for computing and maintaining discrete attributes of geometric objects moving continuously along predefined trajectories. Unlike classical data structures, KDSs maintain not only the desired attribute, but also a set of certificates to prove the validity of said attribute. Furthermore, KDSs provide an update algorithm which preserves the correctness of both the structure and the proof when one of its certificates fails. The effectiveness of a KDS is measured using four properties: efficiency, locality, compactness, and responsiveness. In particular, responsiveness pertains to the speed of the update algorithm: a KDS is considered responsive if it can be updated in polylogarithmic time after a certificate failure. Since their introduction, the responsiveness of KDSs has been analyzed largely in terms of single certificate failures. However, this places a restriction requiring a total ordering of the certificates as they fail. In this paper, we remove this restriction by developing an approach to analyze responsiveness when any number of certificates are allowed to fail at once, and propose an extension of this property that we call batch-responsiveness. We develop batch-responsive update algorithms for a number of KDSs, including tournaments, heaps, sorted lists, convex hulls in 2D, and deformable spanners. We conclude that a KDS enjoys several benefits when made batch-responsive. Removing the requirement of a total ordering of certificate failures, for instance, enables us to avoid high-precision (and computationally expensive) arithmetic. Updating multiple failures at once also enables us to avoid computing intermediate steps on the way to a desired time, granting additional efficiency. Furthermore, the ability to advance time arbitrarily gives us a more robust data structure.

Biography: Duru Turkoglu has joined the CDM School of Computing in 2015. His research interests include computational geometry, graph theory, mesh refinement, and dynamization and kinetization techniques. He received his M.S. and Ph.D. degrees in Computer Science from the University of Chicago. Previously, he received B.S. degrees in Computer Engineering and in Mathematics from the Middle East Technical University in Ankara, Turkey.

4/21 – Dr. Hamed Qahri Saremi – DePaul University
Who Will Drop IT Addiction? A Theoretical Integration and an Empirical Investigation

Abstract:  The use of hedonic information technology (IT), such as social networking sites (SNS), is on the rise, so is the addictions to these systems. Considering the adverse consequences of IT addiction, IT discontinuance has been the subject of recent calls for research. Nonetheless, extant literature still lacks an integrated theoretical understanding of the factors that influence discontinuance of an IT addiction. To fill this gap, this study builds on cognitive dissonance and coping theories to develop and empirically test a structural model that delineates the motivational mechanism toward IT discontinuance in the context of SNS addiction. Using the mechanism validated in the structural model, we then develop a typology of users using latent profile analysis, who may or may not intend to discontinue their SNS addiction. The study makes important contributions to the growing research body on IT addiction, and provides important implications for practice toward discontinuance of IT addiction.

Biography:  Prof. Saremi am an assistant professor of Information Systems (IS) at College of Computing and Digital Media (CDM) at DePaul University. His research focuses on different effective and problematic patterns of use of IS and their impacts on performance at both individual (user) and organizational levels. His research works have appeared in a number of IS journals and conferences such as Journal of Management Information Systems, Journal of Strategic Information Systems, Information & Management, Computers & Education, and Expert Systems with Applications. Prof. Saremi has served in different capacities as an editor, a reviewer, a mini-track chair, and a session chair for various IS journals and conferences. For more details about his research and publications, please visit: https://goo.gl/Vp5dSX

4/28 – Marc Rutzen – EnodoScore
(First of two talks this day)
EnodoScore – Real Estate Analytics

Abstract: Enodo Score is a predictive analytics platform for real estate professionals. The platform offers a suite of products that provide users actionable intelligence to make more informed investment decisions. Utilizing a combination of proprietary technology and established data feeds, Enodo Score’s patent-pending machine learning algorithm delivers industry leading insight.

Biography: As Cofounder and CTO of Enodo Score, Marc Rutzen directs the development and implementation of the platform, including front-end design and development, development of data sharing partnerships, beta testing, customer feedback and business development. Marc is a Licensed Managing Broker in the State of Illinois, and earned his Master of Science in Real Estate Development from Columbia University.

4/28 – Luisa F. Polanía – American Family Mutual Insurance Company.
(Second of two talks this day)
A stroll through some physiological and wellness applications of deep learning and signal processing

Abstract: Deep learning is a fast-growing field in artificial intelligence (AI) that allows a computer to represent data as a nested hierarchy of concepts, with each concept defined in relation to simpler concepts. Deep learning constitutes a viable approach to building AI systems that can operate on complicated, real-world environments and is giving machines near-human levels of visual and speech recognition capabilities.

In this talk, I will describe some applications of deep learning to signal processing problems and physiological and wellness applications. I will first describe how restricted Boltzmann machines (RBMs), the building blocks of deep belief networks, can be employed to model the sparsity pattern of signals in a compressed sensing scenario by capturing statistical dependencies between sparse coefficients, which leads to improvement in reconstruction performance. Results of the RBM-based algorithm in reconstructing electrocardiogram signals will be presented.

The second part of this talk will introduce a convolutional neural network (CNN) architecture that predicts skin condition from facial images. More precisely, the CNN is able to recognize the intensity of wrinkles, pores, and spots by automatically learning features from three datasets captured in Singapore, the United States, and the United Kingdom. Each dataset contains around 20000 images of women between 30 and 45 years old. The third part of this talk will present future research directions in the application of deep learning to unobtrusive physiological monitoring. There has been a tremendous progress in this area motivated by the explosion of the wearable device industry, which allow users to track their vital signals without interrupting their daily life activities. However, this new trend also brings new challenges and redefines the way researchers process physiological signals. For example, new algorithms need to efficiently exploit the massive amount of data that is sensed continuously and use data fusion to combine information from different sensors.

Biography: Luisa F. Polanía received the B.Sc. degree in Electronics Engineering (with honors) from the National University of Colombia in 2009. She received the Ph.D. degree from the Department of Electrical and Computer Engineering at the University of Delaware in 2015. She was a postdoctoral research scientist at the Palo Alto Research Center (PARC) from October 2014 to July 2016. She is currently a machine learning scientist at American Family Mutual Insurance Company. Her research interests include machine learning, deep learning and signal/image/video processing. She was the recipient of the Graduate Student Fellowship in 2013 and the PARC Exceptional Performance Award in 2015. She is the author of 15 publications in top conferences and journals in the areas of signal processing and biomedical engineering. She also holds seven pending U.S. patents.

5/5 – Dr. Vijay K. Gurbani – Bell Labs
Title: Mitigating Mimicry Attacks Against the Session Initiation Protocol

Abstract: The U.S. National Academies of Science’s Board on Science, Technology and Economic Policy estimates that the Internet and voice-over-IP (VoIP) communications infrastructure generates 10% of U.S. economic growth. As market forces move increasingly towards Internet and VoIP communications, there is proportional increase in telephony denial of service (TDoS) attacks. Like denial of service (DoS) attacks, TDoS attacks seek to disrupt business and commerce by directing a flood of anomalous traffic towards key communication servers. In this work, we focus on a new class of anomalous traffic that exhibits a mimicry TDoS attack. Such an attack can be launched by crafting malformed messages with small changes from normal ones.We show that such malicious messages easily bypass intrusion detection systems (IDS) and degrade the goodput of the server drastically by forcing it to parse the message looking for the needed token. Our approach is not to parse at all; instead, we use multiple classifier systems (MCS) to exploit the strength of multiple learners to predict the true class of a message with high probability (98.50% <= p <= 99.12%). We proceed systematically by first formulating
an optimization problem of picking the minimum number of classifiers such that their combination yields the optimal classification performance.  Next, we analytically bound the maximum performance of such a system and empirically demonstrate that it is possible to attain close to the maximum theoretical performance across varied datasets. Finally, guided by our analysis we construct an MCS appliance that demonstrates superior classification accuracy with O(1) runtime complexity across varied datasets.

Biography: Vijay K. Gurbani is a Distinguished Member of Technical Staff at Bell Laboratories’ End-to-End Mobile Network Research department in Nokia Networks. He holds a B.Sc. in Computer Science with a minor in Mathematics, a M.Sc. in Computer Science, both from Bradley University; and a Ph.D. in Computer Science from Illinois Institute of Technology. His current work is focused on scalable analytic architectures and algorithms for autonomic 5G networks. His research interests are multimedia protocols, security and privacy in multimedia protocols, peer-to-peer networks, distributed programming and open/inner source. Vijay’s research has resulted in products that are used in national and international service provider networks. He has over 60 publications in peer-reviewed conferences and journals, 5 books, 7 granted U.S. patents and 18 Internet Engineering Task Force (IETF) RFCs.

5/12 – Dr. David M. Liebovitz, MD, FACP – The University of Chicago
Steps Toward Achieving the Promise of Electronic Health Records
*Biomedical and Health Informatics Workshop Keynote Speaker

Dr. Liebovitz is a general internist committed to optimizing health for individuals and populations of patients, addressing both routine care and chronic disease management. In addition to maintaining his internal medicine clinical practice, he serves as the University of Chicago Medicine’s chief medical information officer.

Dr. Liebovitz is a clinician-educator and informaticist board certified in the new specialty of clinical informatics. His work includes teaching medical students, residents, and graduate school students on topics related to internal medicine, informatics, and patient safety.

His areas of research include exploring how best to leverage information systems to care for patients in the safest manner possible and how next steps in care may be better anticipated and supported in a patient-centric and personalized way. His work has included approaches to enhance system usability, automate learning strategies to safeguard patient privacy, and provide decision support tools for medication safety, laboratory follow-up, and problem list optimization. His research focuses on improving electronic medical records with regard to patient safety, health care quality, and application usability.

Dr. Liebovitz received his electrical engineering degree from the University of Illinois at Urbana-Champaign and earned an MD from the University of Illinois at Chicago. He completed his internal medicine residency at the University of Chicago and served as chief resident at Weiss Memorial Hospital, after which he held a faculty appointment at the University of Chicago until 2002. He then held a variety of operational, research, and educational responsibilities at Northwestern University, focusing on informatics as its chief medical information executive. He also served as site principal investigator for informatics privacy and security related grants sponsored by the National Science Foundation and the Department of Health and Human Services.

5/19 – Dr. Cynthia Putnam – DePaul University
Principles of Accessibility for People with Disabilities and Aging Populations

Abstract: As information and communication technologies become more diffuse, developers and designers need to consider a growing diversity of users including people with disabilities and aging populations. While there is a recognized need to include principles of accessibility in computing education among industry leaders, law makers and academics, there is a lack of shared resources for instructors with little or no background in the area to help them incorporate these topics into their curricula. In response, we created a set of short biographical films that shadow people who have a disability that demonstrate challenges that might be ameliorated with more inclusive considerations in design, including the design of technologies.

People attending the colloquium (May 19) will be asked to participate in a study examining major takeaway’s from the films. Participants will be asked to complete a short questionnaire prior and after viewing the biographical films. (We have also created an online version of the study). Participation in the study is completely voluntary and will not affect students’ standing in the colloquium in any way.

Biography: Cynthia Putnam is an associate professor at the College of Computing and Digital Media at DePaul University. She is interested in human-centered approaches to design. Her framework for this field is informed by over ten years of working in industry as a visual and interaction designer, illustrator and animator prior to earning her PhD. Cynthia received her PhD in Human Centered Design & engineering from the University of Washington.

5/26 – Dr. Clare Bates Congdon – Bowdoin College
*SOCRS Keynote Speaker, DePaul Center, Suite 8005*
It’s not Junk: Using Evolutionary Computation to Infer Functional Regions in Noncoding DNA

Abstract: Over 95 percent of our DNA does not directly code for genes. Formerly called “junk” DNA, it is now understood that some of these noncoding regions are functional…  but determining which regions are functional is a complex biological problem. Computationally, we are able to identify promising candidates for biological investigation, thus hastening the process of scientific discovery. Using the heuristic that motifs (patterns) in noncoding DNA that have been conserved across evolutionary time are more likely to be functional, we search for highly conserved motifs using an evolutionary computation approach. In this talk, I will describe the biological task, the evolutionary computation approach we developed, and results from a recent project that illustrate our success with this work.

Biography: Clare Bates Congdon is a Visiting Associate Professor of Computer Science at Bowdoin College in Brunswick, ME. She works in Evolutionary Computation approaches to Bioinformatics, including inference of regulatory elements in noncoding DNA, phylogenetics, and data mining to investigate gene-environment interactions. Congdon is a former Chair of the IEEE CIS Technical Committee on Bioinformatics and Bioengineering, an Associate Editor of IEEE Transactions on Evolutionary Computation, and a steering committee member of the joint ACM/IEEE Transactions on Computational Biology and Bioinformatics. Her research has been funded by both the National Science Foundation and the National Institutes of Health. Congdon holds a BA from Wesleyan University and an MS and PhD from the University of Michigan

6/2 – Dr. Yolanda Rankin – Spelman College
In-Game Social Interactions that Facilitate Second Language Acquisition

Abstract: In comparison to self-paced tutorials, commercial videogames create an immersive learning environment that allows students to experience the virtual world through sight, sound, participation and imagination. Research shows that one particular genre of videogames, Massively Multiplayer Online Role Playing Games (MMORPGs), provides opportunities for native and non-native speakers to interact with one another during gameplay. Gameplay experiences in MMORPGs are such that language is a means for accomplishing the game tasks and contributes to Second Language Acquisition (SLA), one’s ability to read, write, listen and speak in a targeted language. In contrast, educational videogames intentionally designed for SLA, tend to lack the element of fun or entertainment necessary to sustain learning over time. I posit that we can learn from players’ experiences in both commercial and educational videogames to understand which design elements and aspects of gameplay actually facilitate SLA. Consequently, I examine game dialogue between native English speakers and English as Second Language (ESL) in the popular MMORPG EverQuest2 to understand how these social interactions facilitate ESL students’ proficiency. Next, I examine the gameplay experiences of African American women playing an educational videogame to identify design elements that create inclusive learning environments for diverse players.  Finally, I talk about the role of in-game social interactions as a framework for designing conversational game characters to promote students’ proficiency in the targeted language.

Biography: Yolanda is currently an Assistant Professor in the department of Computer & Information Sciences at Spelman College.  Her research interests include identifying best practices and pedagogical strategies for Computer Science (CS) education K-16, utilizing videogames to promote Second Language Acquisition (SLA), and engaging members of underserved population in participatory design of information-based technologies. A 2016 Woodrow Wilson Career Enhancement Fellow, Yolanda explores in-game social interactions that feature diverse, conversational game characters who facilitate foreign language students’ SLA. Yolanda co-edited the book Moving Students of Color from Consumers to Producers of Technology, which emphasizes strategies for broadening participation of underrepresented groups in CS.

Yolanda accumulated more than twelve years of industry experience before choosing to pursue a career in academia. She developed software applications to support geographically distributed employees who provide customer service during her tenure at IBM Research Lab – Almaden.  Additionally, Yolanda worked as a software engineer and customer technical advocate to develop and deploy wireless features and applications for multiple customers at Lucent Technologies Bell Labs.

Yolanda completed her Ph.D. in Computer Science at Northwestern University as both a National Science Foundation (NSF) Graduate Research Fellow and NSF Alliances for Graduate Education and the Professoriate Fellow. As a Patricia Roberts Harris Fellow, she attained her M.A. in Computer Science at Kent State University and graduated with a B.S. in Mathematics at Tougaloo College, a Historically Black College in Jackson, MS.