2017 Prospective Student Visit Day and Graduate Research Symposium

Ryan Brummet presenting at 2017 UI CS Graduate Research SymposiumThe Computer Science (CS) department welcomed prospective students interested in our graduate pro-grams to our annual Prospective Student Visit Day on Fri, Nov 10. We are looking for strong students with diverse backgrounds to join our MCS and PhD programs. A wide variety of research areas is represented by our world-class faculty including algorithms, computational epidemiology, distributed computing, human-computer interaction, machine learning, massive data algorithms and technology, mobile computing, networks, programming languages, text mining, security, and virtual reality.

An important part of the prospective Student Visit Day is the 3rd Iowa Computer Science Graduate Research Symposium (2017), in which senior PhD students presented talks showcasing their current research. A "keynote" by a CS faculty member followed. Talks were intended for a wide audience with interest in CS, including CS juniors and seniors. The talks presented by current CS graduate students at the Symposium are excellent examples of the exciting CS research taking place here on the UI campus!

Schedule

Friday, Nov 10, 2017

Morning Sessions (in MacLean Hall) for Visiting/Prospective Students

9:30-10:30am

Overview of Graduate Programs, Sriram Pemmaraju, Director of Graduate Studies

10:30-12noon

CS Faculty Meetings and Research Demos

12noon-1:30

Visiting students will meet and have lunch with Computer Science Graduate Students
 

Graduate Student Research Symposium (in Meeting Room A-Iowa City Public Library)

1:35-2:35pm

Session One

1:35pm

Yi Xu

New Optimization Algorithms for Big Data Analytics

1:55pm

Michael Lash

Data, Decisions, and Machine Learning: Optimizing Outcomes via Inverse Classification

2:15pm

Ryan Brummet

Flexible Retransmission Scheduling For Industrial Wireless Sensor Actuator Networks

2:35-2:45pm

Break

2:45-3:45pm

Session Two

2:45pm

Tanmay Inamdar

Capacitated Set Cover using Balls

3:05pm

Huyen Le​

Revisiting The American Voter on Twitter

3:25pm

Kyle Diederich​

Dimensions of Design Approaches for Face-to-Face Collaboration Systems

3:45-4:00pm Break
4:00-5:00pm

Keynote Speaker Omar Chowdhury​

Detecting Security Vulnerabilities in the Modern Age: Formal Techniques To The Rescue

5:00-5:30pm

Reception

Speakers

Yi Xu - UI CS PhD candidateYi Xu

Title: New Optimization Algorithms for Big Data Analytics​

Abstract: The scale and dimensionality of data associated with machine learning and data mining applications (e.g., bioinformatics, finance, social network, computer vision) have seen unprecedented growth in recent years. Classical optimization algorithms are not designed to scale to instances of this size; thus it becomes very important to develop efficient and effective algorithms to solve large-scale problems. In this talk, I will present our recently proposed stochastic optimization algorithms that not only improve theoretical convergence rates but also speed up the performances in real applications.

3rd year PhD student | Advisor: Tianbao Yang | Area of research: Machine Learning


Michael Lash - UI CS PhD CandidateMichael Lash

Title: Data, Decisions, and Machine Learning: Optimizing Outcomes via Inverse Classification

Abstract: “Big data”, “Analytics”, and “Machine Learning” are terms that, in recent years, have been catapulted into the main stream as businesses have adopted the use of of the underlying technologies such terms represent. Consumers, meanwhile, have realized the personal benefit such technologies provide. Inverse classification is a data driven, machine learning technology aimed at providing person-specific plans that lead to optimal outcomes. Such technology works by using a trained machine learning model to guide an instance (e.g., a person) to some desired outcome  (e.g., a desired class grade) by perturbing their features (e.g., time spent studying, time spent with friends, etc.). Furthermore, person-specific factors such as perceived effort and a desire to achieve such an outcome are factored into the optimization.

4th year PhD student | Advisor: Nick Street | Area of research: Machine Learning, Data Mining, Predictive Analytics


Sayan Bandyapadhyay - UI CS PhD CandidateRyan Brummet

Title: Flexible Retransmission Scheduling For Industrial Wireless Sensor Actuator Networks​

Abstract: Real-time and reliable communication is essential for industrial wireless sensor actuator networks. To this end, researchers have proposed a wide range of TSCH retransmission scheduling techniques. However, these methods usually employ a link centric scheduling approach which allocates a fixed number of retransmissions for each link of a flow. The lack of flexibility of this approach is problematic because failures do not occur uniformly across links and link quality changes over time. Instead, we present flow centric scheduling, a scheduling approach that flexibly and dynamically reallocates retransmissions among the links traversed by a flow at run-time. Simulation results show that under flow centric scheduling the median real-time capacity increases as much as 64%. This result is due to the flexibility of flow centric scheduling which enables it to use fewer retransmissions per flow and cope with non-uniform and variable link quality better than link centric scheduling.

5th year PhD student | Advisor: Octav Chipara | Area of research: Wireless Sensor Networks


Tanmay Inamdar - UI CS PhD CandidateTanmay Inamdar

Title: Capacitated Set Cover using Balls

Abstract: In this work, we study a version of Geometric Set Cover with capacities. The input consists of points and balls in a metric space. The input also consists of an integer U, called the capacity, which is the maximum number of points a ball is allowed to cover. The objective is to choose the minimum number of balls such that each point is covered, while respecting the capacities of the balls. Since this problem is a generalization of the well-known Minimum Set Cover problem, it is NP-hard to approximate it within a factor of o(log n). We use Linear Programming to obtain an (α, β)-bicriteria approximation for this problem, where α and β are constants. That is, we obtain a solution where the balls may be expanded by a factor of β, and the cost of the solution is at most α times that of the optimal solution (that is not allowed to expand the balls).

3rd year PhD student | Advisor: Kasturi R. Varadarajan | Area of research: Approximation Algorithms, Computational Geometry, and Distributed Algorithms


Huyen Le - UI CS PhD candidateHuyen Le

TitleRevisiting The American Voter on Twitter

Abstract: The American Voter - a seminal work in political science - uncovered the multifaceted nature of voting behavior which has been corroborated in electoral research for decades since. In this paper, we leverage The American Voter as an analysis framework in the realm of computational political science, employing the factors of party, personality, and policy to structure the analysis of public discourse on online social media during the 2016 U.S. presidential primaries. Our analysis of 50 million tweets reveals the continuing importance of these three factors; our understanding is also enriched by the application of sentiment analysis techniques. The overwhelmingly negative sentiment of conversations surrounding 10 major presidential candidates reveals more "crosstalk'' from Democratic leaning users towards Republican candidates, and less vice-versa. We uncover the lack of moderation as the most discussed personality dimension during this campaign season, as the political field becomes more extreme - Clinton and Rubio are perceived as moderate, while Trump, Sanders, and Cruz are not. While the most discussed issues are foreign policy and immigration, Republicans tweet more about abortion than Democrats who tweet more about gay rights than Republicans. Finally, we illustrate the importance of multifaceted political discourse analysis by applying regression to quantify the impact of party, personality, and policy on national polls.

5th year PhD student | Advisor: Zubair Shafiq | Area of research: Social Media Analysis, Text Mining, Applied Machine Learning


Kyle Diederich - UI CS PhD CandidateKyle Diederich

Title: Dimensions of Design Approaches for Face-to-Face Collaboration Systems​

Abstract: In this presentation, we will examine current design approaches used during the creation of computing technologies aimed at young children. We will focus our exploration along multiple dimensions, including user tracking, user proximity, attention to technology, and modes of interaction with technological systems. This analysis will discuss the focus of the human-computer interaction community over time. We will then argue for an increase in efforts to support design approaches which facilitate inclusion of young children in the design of new technologies.

3rd year PhD student | Advisor: Juan Pablo Hourcade | Area of research: Human-Computer Interaction


Omar Chowdhury - Assistant Professor of Computer Science at the University of Iowa.

 

Keynote Speaker Omar Chowdhury

Title: Detecting Security Vulnerabilities in the Modern Age: Formal Techniques To The Rescue 

Abstract: The common perception that security threats a modern critical system or infrastructure have to face only come from a lone hacker is constantly being challenged. Emerging threats such as surveillance and censorship from resourceful adversaries such as nation states and intelligence agencies have become a real concern in the modern age. More often than not, these influential adversaries exploit undetected security vulnerabilities to circumvent the possible security measures employed by a critical system. It is hence paramount to design and develop highly automated approaches for detecting possible vulnerabilities.

A majority of the prior efforts invested on leveraging and enhancing automated software testing approaches (e.g., unguided black-box fuzzing) for detecting low-level memory errors (e.g., buffer overflow)  whereas logical vulnerabilities abound in practice. Existing work has often avoided leveraging formal techniques (e.g., model checking, SMT solving) due to the impression that leveraging these techniques either would induce scalability challenges or would require significant manual effort. The speaker will provide two recent results---one in the context of secure web communication and another in the context of telecommunication network---where formal techniques have been very effective in finding intricate logical vulnerabilities which can be exploited by adversaries to violate inherent security guarantees.

Bio: Dr. Omar Haider Chowdhury is an Assistant Professor of Computer Science at the University of Iowa. Dr. Chowdhury's research focus is on leveraging formal machinery and techniques to solve practically-relevant security and privacy problems of emerging systems. Dr. Chowdhury currently co-directs the Computational Logic Center (CLC), and is also an active member of the Informatics Initiative (UI3). Before joining the University of Iowa, he was a post-doc at Carnegie Mellon University and Purdue University. He received his Ph.D. in Computer Science from the University of Texas at San Antonio.


 

Directions

Getting to the Conference

Iowa City lies just off of Interstate 80 in Eastern Iowa. Regardless of whether you are coming from the West or the East, you will want to take exit 244 off of I-80 following Dubuque Street turning right on Washington Street, then left on Linn St. The Dubuque St Parking Ramp will be on your right, just past the Iowa City Public Library. For a visual representation and additional parking options, see the map below, courtesy of our 2017 host.

Parking map of Iowa City Public Library