DISSERTATION DEFENSE Department of Computer Science and Engineering, University of South Carolina Candidate: Yuewei Lin Advisor: Dr. Song Wang Date: March 16, 2016 Time: 12:00 P.M. Place: Swearingen 3A75 Abstract Interest detection is to detect an object, event, or process which causes attention. In this work, we focus on the interest detection in the image, video and multiple videos. Interest detection in an image or a video, is closely related to the visual attention. However, the interest detection in multiple videos needs to consider all the videos as a whole rather than considering the attention in each single video independently. In this work, we first introduce a new computational visual-attention model for detecting region of interest in static images and/or videos. This model constructs the saliency map for each image and takes the region with the highest salient value as the region of interest. Specifically, we use the Earth Mover's Distance (EMD) to measure the center-surround difference in the receptive field. Furthermore, we propose to take two steps of biologically-inspired nonlinear operations for combining different features. Then, we extend the proposed model to construct dynamic saliency maps from videos, by computing the center-surround difference in the spatio-temporal receptive field. Motivated by the natural relation between visual saliency and object/region of interest, we then propose an algorithm to detect infrequently moving foreground, in which the saliency detection technique is used to identify the foreground (object/region of interest) and background. Finally, we focus on the task of locating the co-interest person from multiple temporally synchronized videos taken by the multiple wearable cameras. More specifically, we propose a co-interest detection algorithm that can find persons that draw attention from most camera wearers, even if multiple similar-appearance persons are present in the videos. We built a Conditional Random Field (CRF) to achieve this goal, by taking each frame as a node and the detected persons as the states at each node.

More info at: https://www.facebook.com/events/943148319099082/

Stephen Fenner, University of South Carolina Abstract: We show that the bipartite perfect matching problem is in quasi-NC. In particular, it has uniform circuits of quasi-polynomial size and O(log^2 n) depth. Previously, only an exponential upper bound was known on the size of such circuits with poly-logarithmic depth. We obtain our result by an almost complete derandomization of the Isolation Lemma of Mulmuley, Vazirani, & Vazirani, which was used to yield an efficient randomized parallel algorithm for the bipartite perfect matching problem. Time permitting, we describe an RNC algorithm to find a perfect matching in a bipartite graph using O(log^2 n) random bits. Bio: Stephen Fenner is a professor of Computer Science and Engineering at the University of South Carolina. His research interests are in theoretical computer science and include computational complexity, computability, algorithms, and quantum informatics. Be aware, this talk will be held in an earlier timeslot (1:45 - 3:00) in a different room (2A19) than usual. This seminar is open to anyone who is interested, not just students enrolled in the CSCE 791 class. Please consider attending.

Duncan Buell, University of South Carolina Abstract: The pedagogical center of many university First Year Composition programs is the revision of essays, the notion that the draft of a student’s English 101 essay should be revised before being turned in as a final version. Most of the established research on FYC concludes that students revise in a shallow way, correcting minor grammatical errors and doing minor word substitution. This research has, however, been conducted by human beings examining small sets of FYC essays. Dr. Buell, together with Dr. Chris Holcomb, Director of First Year English at USC, have been looking at revision in the ENGL 101/102 as a “big data” computation. They have written Python code to compare draft and final versions on specific and targeted features that can be examined by computer. Based on an early corpus of 439 papers from 2014-2015, it would seem that the established conclusions about student revision are just wrong, and that student revision is much different thing. Buell and Holcomb, with a team of graduate and undergraduate students funded by the Center for Digital Humanities, are working to collect all 10,000 (plus or minus) essays from ENGL 101 and 102 in the spring 2016 and following semesters, and to process them all to examine revision and writing characteristics. This is thus a combination of a “big data” and a “natural language” computation. We emphasize that although we use natural language packages, this is not software to “grade” or “assess” the writing. Rather, we have targeted characteristics thought to be typical of student (and compared against “academic”) writing, and we are computing quantified measurements of these characteristics. Bio: Duncan A. Buell is a Professor in the Department of Computer Science and Engineering at the Unviversity of South Carolina. His Ph.D. is in mathematics from the University of Illinois at Chicago (1976). He was from 2000 to 2009 the department chair at USC, and in 2005-2006 was interim dean. He has done research in document retrieval, computational number theory, and parallel computing, and has more recently turned to digital humanities as one of the emerging “marketplace” applications for computing. He is engaged with First Year English at USC on the analysis of freshman English essays, searching for an understanding of actual student writing in an effort to improve pedagogy for first year English instruction. He has team taught four times with Dr. Heidi Rae Cooley on the presentation of unacknowledged history on mobile devices, and he and Dr. Cooley are actively engaged in ways to go beyond text to fully enable the use of visual media in mobile applications that present humanities content, especially content that might normally remain unacknowledged by institutional authority. This seminar is open to anyone who is interested, not just students enrolled in the CSCE 791 class. Please consider attending.

Matthew Brashears, University of South Carolina (Department of Sociology) Abstract: Humans make mistakes but diffusion through social networks is typically modeled as though they do not. We find in an experiment that high entropy message formats (text messaging pidgin) are more prone to error than lower entropy formats (standard English). We also find that efforts to correct mistakes are effective, but generate more mutant forms of the contagion than would result from a lack of correction. This indicates that the ability of messages to cross “small-world” human social networks may be overestimated and that failed error corrections create new versions of a contagion that diffuse in competition with the original. Bio: Matthew E. Brashears is an Associate Professor of Sociology at the University of South Carolina. His current research focuses on linking cognition to social network structure, studying the effects of error and error correction on diffusion dynamics, and using ecological models to connect individual behavior to collective dynamics. His work has appeared in Nature Scientific Reports, the American Sociological Review, Social Networks, Sociological Science, and Social Psychology Quarterly, among others. He has received grants from the National Science Foundation, the Defense Threat Reduction Agency, and the Army Research Office. He currently serves on the editorial board for Social Psychology Q This seminar is open to anyone who is interested, not just students enrolled in the CSCE 791 class. Please consider attending.

Gregory Dudak, McGill University Abstract: This talk will address the deployment of robotic systems for data collection. This includes task specification, gait learning and data analysis. As a concrete example I will discuss the automated analysis of video data, and specifically video data collected underwater with an amphibious vehicle (the Aqua 2 hexapod). Automated systems can collect data at prodigious rates and the timely analysis of this data is a growing challenge, especially when there are bandwidth constraints between the data source and the people who must examine the data. We are specifically interested in the real-time summarization and detection of the most interesting events in a video sequence, for use by humans who will analyze the data either in real time, or offline. To do this, we are developing methods that adapt to video data streams in real time to collect salient events and using them in the context of a group of vehicles that fly, swim and float. Bio: Gregory Dudek is the Director of the School of Computer Science, a James McGill Professor, member of the McGill Research Centre for Intelligent Machines (CIM) and an Associate member of the Dept. of Electrical Engineering at McGill University. He is the former Director of McGill's Research Center for Intelligent Machines, a 25 year old inter-faculty research facility. In 2010 he was awarded the Fessenden Professorship in Science Innovation and also received the prix J. Armand Bombardier for Technological Innovation Robotics from ACFAS, the Association francophone pour le savoir (the French learned society). He is also the recipient of the Canadian Image Processing and Pattern Recognition Award for Research Excellence and the award for Service to the Community at the Conference on Computer and Robot Vision. He directs the McGill Mobile Robotics Laboratory. He has been on the organizing and/or program committees of Robotics: Systems and Science, the IEEE International Conference on Robotics and Automation (ICRA), the IEEE/RSJ International Conference on Intelligent Robotics and Systems (IROS), the International Joint Conference on Artificial Intelligence (IJCAI), Computer and Robot Vision, IEEE International Conference on Mechatronics and International Conference on Hands-on Intelligent Mechatronics and Automation among other bodies. He is president of CIPPRS, the Canadian Information Processing and Pattern Recognition Society, an ICPR national affiliate. He was on leave in 2000-2001 as Visiting Associate Professor at the Department of Computer Science at Stanford University and at Xerox Palo Alto Research Center (PARC). During his sabbatical in 2007-2008 he visited the Massachusetts Institute of technology and co-founded the company Independent Robotics Inc. He obtained his PhD in computer science (computational vision) from the University of Toronto, his MSc in computer science (systems) at the University of Toronto and his BSc in computer science and physics at Queen's University. He has published over 200 research papers on subjects including visual object description and recognition, robotic navigation and map construction, distributed system design and biological perception. This includes a book entitled "Computational Principles of Mobile Robotics" co-authored with Michael Jenkin and published by Cambridge University Press. He has chaired and been otherwise involved in numerous national and international conferences and professional activities concerned with Robotics, Machine Sensing and Computer Vision. His research interests include perception for mobile robotics, navigation and position estimation, environment and shape modelling, computational vision and collaborative filtering. This seminar is open to anyone who is interested, not just students enrolled in the CSCE 791 class. Please consider attending.

TECH TUESDAY TALKS Presented by TecHealth-Technology Center to Promote Healthy Lifestyles A South Carolina SmartState Center Date: Tuesday, February 9 Time: Noon Place: Discovery 1, 915 Greene Street, Room 140 Speaker: Dr. Jenay M. Beer Dr. Beer is an engineering psychologist specializing in human-robot interaction (HRI) for the older adult population. Primary research interests include the application of technology to improve the quality of lives for older adults, as well as the application of assistive technology for older individuals with disabilities. Assistant Professor-joint position in USC Computer Science and Engineering Department & the College of Social Work · Director of the Assistive Robotics and Technology Lab · Associate Director of Usability for SmartHOME Research Initiative Tech Tuesday Talks, is a new, monthly seminar series bringing together researchers from across the USC campus who share interest in technology-assisted health promotion and disease prevention interventions and research. The series presents a forum to learn about one another’s work, spark collaborations, as well as to introduce students to the ongoing research conducted on the USC campus which incorporates technology in health promotion. All interested faculty, staff, students and the general public are invited to attend. Tech Tuesday Talks are presented on the second Tuesday of the month in the Discovery I Building, Room 140 at noon. To learn more, visit http://techealth.sc.edu/tech-tuesday-talks

Nicholas Stiffler, University of South Carolina Abstract: The speed at which robots begin to enter various application domains is now largely dependent on the availability of robust and efficient algorithms that are capable of solving the complex planning problems inherent to the given domain. This talk will present a line of research that makes progress towards solving some of the complex planning problems found in target tracking applications where a robot or team of robots seeks to locate and follow a group of moving targets. First, I will describe an algorithm for computing the optimal searcher motion strategy when there exists just a single searcher. Second, I will discuss the multiple searcher scenario and present a deterministic and sampling-based algorithm that coordinates the motion of the searchers. The overall theme is that the design and implementation of robust and efficient planners is imperative for robots to manage the complex tasks we envision for them. Bio: Nicholas Stiffler is a Ph.D. candidate in the department of Computer Science and Engineering at the University of South Carolina. His research focuses on on the design and implementation of planners for a variety of complex robotic planning problems. He received his M.S. (2012) and B.S. (2009) degrees from the University of South Carolina. This seminar is open to anyone who is interested, not just students enrolled in the CSCE 791 class. Please consider attending.