All Seminars

• Upcoming Seminars
• All Seminars

Title: Navigating a maze differently - a dissection of human spatial decision making in Immersive Virtual Reality
Seminar: Computer Science
Speaker: Dr. Arya Basu, Emory University
Contact: Vaidy Sunderam, VSS@emory.edu
Date: 2021-09-24 at 1:00PM
Venue: https://emory.zoom.us/j/98352727203
Download Flyer  Add to Calendar
Abstract: Navigating spaces is an embodied experience. Examples can vary from rescue workers trying to save people from natural disasters, tourists finding their way to the nearest coffee shop, or a person solving a maze. Virtual reality can allow these experiences to simulate in a controlled virtual environment. However, virtual reality users remain anchored in the real world and the conventions by which the virtual environment gets deployed influence user performance. There is currently a need to evaluate the degree of influence imposed by extrinsic factors and virtual reality hardware on its users. Traditionally, immersive virtual reality experiences uses Head-Mounted Displays with powerful computers rendering the virtual environment's graphical content. However, user input has been facilitated using various human interface devices, including Keyboards, Mice, Trackballs, Touchscreens, Joysticks, Gamepads, Motion detecting cameras, and Webcams. Some of these HIDs have also been introduced for non-immersive video games and general computing. Thus, a subset of virtual reality users has greater familiarity than others in using these HIDs. Virtual reality experiences that utilize gamepads (controllers) to navigate virtual environments may introduce a bias towards usability among virtual reality users previously exposed to video gaming. For widespread adoption, we must offer generalizable interaction paradigms to users of all shapes and forms, including video-gamers, non-video-gamers, and everyone else. To establish universality, the field must first understand the different ways users engage and perceive virtual spaces. In this talk, we will be observing an evaluative user study conducted using our ubiquitous virtual reality framework with general audiences. Among our findings, we reveal a usability bias among virtual reality users who are predominantly video gamers. Beyond this, we will overview a recent work on dissecting deeper into the human spatial trajectory and uncovering a set of observed dynamic parameters affecting user performance. Lastly, we will end this talk by briefly summarizing plans to explore Long short-term memory (LSTM) neural networks to investigate human spatial trajectories' hidden parameters that affect their spatial decision-making capability. Biography: Aryabrata Basu is the Staff Research Scientist at the Emory Center for Digital Scholarship (ECDS). Basu creates, modifies, and configures 3D models using various computer modeling, simulation software, and geospatial data in this capacity. In partnership with faculty and ECDS staff, Basu prepares aesthetically composed digital media through graphic design, image processing, and data visualization for use in ECDS-supported digital scholarship projects. Basu primarily researches and explores new methods of visualizing data platforms, including virtual reality, augmented reality, and mixed reality.

Title: Trustworthy Machine Learning: From Theory to Practice
Seminar: Computer Science
Speaker: Dr. Yuan Hong, IIT, Illinois Institute of Technology
Contact: Dr. Li Xiong, lxiong@emory.edu
Date: 2021-09-17 at 1:00PM
Venue: https://emory.zoom.us/j/98352727203
Download Flyer  Add to Calendar
Abstract: Machine learning has achieved many big innovations in all industries and significantly impacted our daily lives. However, machine learning can also result in severe security and privacy risks. In this talk, I will present our recent works on both theory and applications that fundamentally contribute to machine learning security and privacy. First, we designed the first differential privacy mechanism (R2DP) that universally optimizes the randomization for the maximum utility w.r.t. any utility metric. It fundamentally improves the utility of differential privacy mechanisms in all the relevant applications, such as statistical queries, classification, social network analysis, and deep learning. Second, we propose the first black-box attack framework that generates universal 3-dimensional (U3D) perturbations to subvert a wide variety of video deep neural networks (DNNs). The new attack is easy-to-launch, universal, transferable, and human-imperceptible. It can also bypass the state-of-the-art defense methods. Such novel attack motivates the video recognition systems to build and integrate more robust DNN models. Biography: Yuan Hong is an Assistant Professor of Computer Science and Cybersecurity Program Director at Illinois Institute of Technology. He received his Ph.D. degree from Rutgers University in 2014. His research interests primarily lie in the fields of security, privacy, optimization, and data science, such as differential privacy, secure multiparty computation, applied cryptography, adversarial learning, and certified robustness. He is a recipient of the NSF CAREER award, and his work has appeared in prestigious security and data science venues such as Oakland, CCS, PETS, AAMAS, CIKM, EDBT, ICDCS, TDSC, TKDE, TIFS, and TOPS. His research is supported by multiple NSF and AFOSR awards.

Title: Toward Robust Abstractive Multi-Document Summarization and Information Consolidation
Seminar: Computer Science
Speaker: Dr. Fei Liu, UCF
Contact: Jinho Choi, jinho.choi@emory.edu
Date: 2021-09-10 at 1:00PM
Venue: https://emory.zoom.us/j/98352727203
Download Flyer  Add to Calendar
Abstract: Humans can consolidate textual information from multiple sources and organize the content into a coherent summary. Can machines be taught to do the same? The most important obstacles facing multi-document summarization include excessive redundancy in source content, less-understood sentence fusion and the looming shortage of training data. In this talk, I will present our recent work tackling these issues through decoupling of content selection and surface realization. I will describe a lightly-supervised optimization framework using determinantal point processes (DPP) for content selection. I will further present a new method leveraging DPP to select self-contained summary segments to be highlighted on the source documents to make it easier for users to navigate through a large amount of text. Finally, I will discuss challenges and opportunities for driving forward research on abstractive multi-document summarization. Biography: Dr. Fei Liu is an associate professor of Computer Science at the University of Central Florida, where she leads the Natural Language Processing Group. Her research areas are natural language processing and machine learning, with a special emphasis on automatic summarization. Her research aims to generate summaries from a massive amount of textual data to combat information overload. Building on recent advances in deep learning, Dr. Liu's research explores both extractive and abstractive approaches to produce informative, succinct and accurate summaries. Dr. Liu was a postdoctoral fellow at Carnegie Mellon University, member of Noah's ARK. She worked as a senior scientist at Bosch Research, Palo Alto, California, one of the largest German companies building intelligent car systems and home appliances. Liu received her Ph.D. in Computer Science from the University of Texas at Dallas, supported by Erik Jonsson Distinguished Research Fellowship. She obtained her Bachelor and Master's degrees in Computer Science from Fudan University. Dr. Liu has published 60+ peer-reviewed papers in leading conferences and journals. She regularly serves on program committees of major international conferences. Liu was selected for the 2015 "MIT Rising Stars in EECS" program. Her work was nominated as Best Paper Award Finalist at WWW 2016 and Area Chair Favorite Paper at COLING 2018.

Title: Automatically Measuring Emotion from Speech: New Methods to Move from the Lab to the Real World
Seminar: Computer Science
Speaker: Dr. Emily Mower Provost, University of Michigan
Contact: Jinho Choi, jinho.choi@emory.edu
Date: 2021-09-03 at 1:00PM
Venue: https://emory.zoom.us/j/98352727203
Download Flyer  Add to Calendar
Abstract: Emotion has intrigued researchers for generations. This fascination has permeated the engineering community, motivating the development of affective computing methods. However, human emotion remains notoriously difficult to accurately detect. As a result, emotion classification techniques are not always effective when deployed. This is a problem because we are missing out on the potential that emotion recognition provides: the opportunity to automatically measure an aspect of behavior that provides critical insight into our health and wellbeing, insight that is not always easily accessible. In this talk, I will discuss our efforts in developing multimodal emotion recognition approaches that are effective in natural environments and demonstrate how these approaches can be used to support mental health. Venue: https://emory.zoom.us/j/98352727203

Title: Welcome to CS700 / CS Seminar Fall 2021
Seminar: Computer Science
Speaker: Dr. Vaidy Sunderam, Emory University
Contact: Vaidy Sunderam, vss@emory.edu
Date: 2021-08-27 at 1:00PM
Venue: https://emory.zoom.us/j/98352727203
Download Flyer  Add to Calendar
Abstract: Our first meeting will be Fri 27 Aug 1-2pm ET, which will feature a town hall hosted by the CSI program admins Dorian Arnold, Matt Reyna, Steve Qin, Shun Yan Cheung, and Yvette Hilaire. Please be sure to block off every Fri 1-2pm ET, but Yvette will also send a calendar invite each week with specific details of that week's seminar. Last but not least, I invite all of you to suggest speakers and/or formats that we should include. Thank you and best wishes Vaidy CS700 Instructor

Title: Crowdsourcing and Semi-Supervised Learning for Detection and Prediction of Hospital Acquired Pressure Ulcer Injury
Defense: Computer Science
Speaker: Mani Sotoodeh, Emory University
Contact: Joyce Ho, joyce.c.ho@emory.edu
Date: 2021-07-27 at 1:00PM
Venue: https://emory.zoom.us/j/93643444080
Download Flyer  Add to Calendar
Abstract: Pressure ulcer injury (PUI) or bedsore is “a localized injury to the skin and/or underlying tissue due to pressure.” More than 2.5 million Americans develop PUI annually, and the incidence of hospital-acquired PUI (HAPUI) is around 5% to 6%. Bedsores are correlated with reduced quality of life, higher mortality and readmission rates, and longer hospital stays. The Center for Medicare and Medicaid considers PUI as the most frequent preventable event, and PUIs are the 2nd most common claim in lawsuits. The current practice of manual quarterly assessments for a day to estimate PUI rates has many disadvantages including high cost, subjectivity, and substantial disagreement among nurses, not to mention missed opportunities to alter practices to improve care instantly. The biggest challenge in HAPUI detection using EHRs is assigning ground truth for HAPUI classification, which requires consideration of multiple clinical criteria from nursing guidelines. However, these criteria do not explicitly map to EHRs data sources. Furthermore, there is no consistent cohort definition among research works tackling HAPUI detection. As labels significantly impact the model’s performance, inconsistent labels complicate the comparison of research works. Multiple opinions for the same HAPUI classification task can remedy this uncertainty in labeling. Research works on learning with multiple uncertain labels are mainly developed for computer vision. Unfortunately, however, acquiring images from PUIs at hospitals is not standard practice, and we have to resort to tabular or time-series data. Finally, acquiring expert nursing annotations for establishing accurate labels is costly. Though if unlabelled samples can be utilized, a combination of annotated and unlabelled samples could yield a robust classifier. To overcome these challenges, we achieved the following: 1) Proposing a new standardized HAPUI cohort definition applicable to EHR data loyal to clinical guidelines; 2) Introducing a novel model for learning with unreliable crowdsourcing labels using sample-specific perturbations, suitable for sparse annotations of HAPUI detection (CrowdTeacher); 3) Exploring unstructured notes for CrowdTeacher enhancement and gleaning better feature representations for HAPUI detection; 4) Incorporating unlabelled data into HAPUI detection via semi-supervised learning to reduce annotation costs

Title: Federated Tensor Factorization for Collaborative Health Data Analytics
Seminar: Computer Science
Speaker: Jing Ma, Emory University
Contact: Joyce Ho, joyce.c.ho@emory.edu
Date: 2021-06-22 at 2:00PM
Venue: https://emory.zoom.us/j/8187241545
Download Flyer  Add to Calendar
Abstract: Modern healthcare systems are collecting a huge volume of healthcare data from a large number of individuals with various medical procedures, medications, diagnosis, lab tests and so on. Tensor factorization has been demonstrated as an efficient approach for computational phenotyping, where massive electronic health records (EHRs) are converted to concise and meaningful clinical concepts. However, the EHR data is also fragmented and is always distributed among independent medical institutions, and they are prohibited from being shared and exchanged. Recently, federated learning offers a paradigm for collaborative learning among different entities, which seemingly provides an ideal potential to further enhance the tensor factorization-based collaborative phenotyping to handle sensitive personal health data. This poses challenges to preserving the privacy of the exchanged intermediary results in order to protect the sensitive patient information. Meanwhile, efforts still need to be made to overcome the limitations of the federated tensor factorization, including the restrictions to the classic tensor factorization, high communication cost and reduced accuracy. Furthermore, it is essential to develop the decentralization techniques for federated tensor factorization to deal with the vulnerability of the central server to malfunction and external attacks. To deal with these challenging problems, we propose 1) a privacy-preserving collaborative tensor factorization method for computational phenotyping which is able to deal with heterogeneous data with rigorous privacy guarantee and achieves less communication cost and comparable accuracy; 2) a communication efficient federated generalized tensor factorization, which is flexible enough to choose from a variate of losses to best suit different types of data in practice; 3) a communication efficient decentralized generalized tensor factorization method which enables the absence of the central server and further reduces the communication cost.

Title: Cops, Robbers, and Barricades
Seminar: Computer Science
Speaker: Erin Meger, Université du Québec à Montréal
Contact: Vaidy Sunderam, vss@emory.edu
Date: 2021-06-14 at 1:00PM
Venue: https://emory.zoom.us/j/93995724437
Download Flyer  Add to Calendar
Abstract: Combinatorial games, especially games on graphs, have become more popular as a research interest over the past few decades. Despite their fun nature, some deep conjectures still remain elusive. In this talk, we will focus on what is now a classic game on graphs, Cops and Robbers. We will discuss the history of the game, some basic results and a few variants. Mostly, we will focus on Cops, Robbers and Barricades, where the Robber is allowed to build vertex-barricades. We will include a rigorous characterization of the graphs where a single cop can win, as well as a complexity result for the problem in general.

Title: The Seeing-eye Robot Grand Challenge: Developing a Human-Aware Artificial Collaborator
Seminar: Computer Science
Speaker: Reuth Mirsky, The University of Texas at Austin
Contact: Vaidy Sunderam, vss@emory.edu
Date: 2021-04-28 at 10:00AM
Venue: https://emory.zoom.us/j/91543803144
Download Flyer  Add to Calendar
Abstract: In this talk I will present the seeing-eye robot grand challenge and discuss the components required to design and build a service robot that can replace or surpass the functionalities of a seeing-eye dog. This challenge encompasses a variety of research problems that can benefit from human-inspired AI: reasoning about other agents, human-robot interactions, explainability, teaching teammates, and more. For each of these problems, I will present an example novel contribution that leverages the bilateral investigation of human and artificial intelligence. Finally, I will discuss the many remaining challenges towards achieving a seeing-eye robot and how I plan to tackle these challenges. Bio: Reuth Mirsky is a Postdoctoral Fellow at the Computer Science Department in the University of Texas as Austin. She received her Ph.D. on plan recognition in real world environments from the Department of Software and Information Systems Engineering in Ben Gurion University. She is interested in the similarities and the differences between AI and natural intelligence, and how these can be used to extend AI. In her research, she seeks algorithms, behaviors and frameworks that can improve existing AI with human-inspired design. Beyond her research, Reuth is an active member in the AI research community. Some of her recent roles are: a chair for the Plan, Activity, and Intent Recognition (PAIR) workshop as part of the AAAI workshop series, a guest editor in Frontiers of Artificial Intelligence in a special issue on Plan and Goal Recognition, a program committee member for AAMAS 2021, and a reviewer for AIJ, JAIR, and RA-L. Reuth was selected as one of the 2020 Electrical Engineering and Computer Science (EECS) Rising Stars. In addition, her work has led to several awards including two awards from the Israeli Ministry of Science (Award for Leading Applied Research and scholarship for Excelling Women in STEM) and the Eric and Wendy Schmidt Postdoctoral Award for Women in Mathematical and Computing Sciences.

Title: Digital Humanities and Computer Science: Intersections and Opportunities
Seminar: Computer Science
Speaker: Lauren Klein, Emory University - QTM
Contact: Jinho Choi, jinho.choi@emory.edu
Date: 2021-04-23 at 1:00PM
Venue: https://emory.zoom.us/j/92103915275
Download Flyer  Add to Calendar
Abstract: Abstract: What is the field of digital humanities? What are its intersections with computer science? And what are the opportunities for collaboration between digital humanists and computer scientists? This talk will provide an overview of the digital humanities, a wide-ranging field that involves the use of computational methods to explore humanistic research questions, and the use of humanistic methods to explore issues related to computation. In order to illustrate this range of work, as well as the ways in which computer scientists in various subfields—particularly NLP/ML and data visualization—can contribute, I will present three recent collaborative research projects: 1) the development of a model of lexical semantic change which, when combined with network analysis, offers a new perspective on the abolitionist movement of the 19th century United States; 2) the design and fabrication of a large-scale haptic data visualization, inspired by a forgotten historical visualization scheme, which suggests future possibilities for visualization design; and 3) a book, Data Feminism, which outlines a set of feminist principles for more just and equitable data science, and computer science more broadly. The goal of this talk is to foster future collaborations between the Digital Humanities Lab and Emory CS faculty and students. Lauren Klein is an associate professor in the departments of English and Quantitative Theory \& Methods at Emory University, where she also directs the Digital Humanities Lab. Before moving to Emory, she taught in the School of Literature, Media, and Communication at Georgia Tech. Klein works at the intersection of digital humanities, data science, and early American literature, with a research focus on issues of gender and race. She is the author of An Archive of Taste: Race and Eating in the Early United States (University of Minnesota Press, 2020) and, with Catherine D’Ignazio, Data Feminism(MIT Press, 2020). With Matthew K. Gold, she edits Debates in the Digital Humanities a hybrid print-digital publication stream that explores debates in the field as they emerge. Her work has appeared in leading humanities journals including PMLA, American Literature, and American Quarterly; and at technical conferences including NACCL, EMNLP, and IEEE VIS. Her work has been supported by grants from the NEH and the Mellon Foundation.