| |
|
|
The Department of Computer Science
of Colorado State University, in cooperation with ISTeC
(Information Science and Technology Center), offers the CS
Colloquium series as a service to all who are interested
in computer science. When in-person meetings are possible, most seminars are scheduled for
Monday 11:00AM -- 11:50AM in CSB 130 or Morgan Library
Event Hall. For help finding the locations of our seminar
meetings, consult the on-line CSU campus map. |
Upcoming Events
CS Colloquium Schedule, Spring 2024
|
|
Computer Science Department Colloquium Introduction to the Graduate Program Speaker: Sanjay Rajopadhye, Professor and Graduate Director, Computer Science Department When: 11:00AM ~ 11:50AM, Monday January 22, 2024  Abstract: Dr. Rajopadhye introduces the Computer Science graduate program at CSU. |
|
|
|
Computer Science Department Colloquium Unmasking Vulnerabilities: The Provocative Dance of Device Physics in Flash Storage Security Speaker: Biswajit Ray, Associate Professor, Electrical and Computer Engineering Department, Colorado State University When: 11:00AM ~ 11:50AM, Monday January 29, 2024 Abstract: Non-volatile memories often become the central focus of cyber-physical attacks, especially in resource-constrained Internet of Things (IoT) applications. While various software-based security protocols and encryption techniques are commonly used in computer security to safeguard sensitive information, recent studies indicate their vulnerability to sophisticated cyber-physical attacks, including cold-boot attacks, row-hammer attacks, and fault-injection attacks. This susceptibility poses a substantial threat, particularly for embedded flash storage applications. In this presentation, I will elucidate how the intricate device physics of memory cells can be used to unveil novel attack surfaces in flash storage. These include the potential for data recovery from deleted memory locations and the unauthorized manipulation of data through the application of ionizing radiation. Intriguingly, I will also demonstrate how an understanding of device physics can be harnessed to fortify the memory system, serving to counter and mitigate the impacts of these cyber-physical attacks. Bio: Biswajit Ray is an Associate Professor of Electrical and Computer Engineering at Colorado State University, where he leads the Reliable and Assured Microelectronics (RAM) Laboratory. Dr. Ray earned his Ph.D. from Purdue University in West Lafayette, IN, and subsequently worked at SanDisk Corporation in Milpitas, California, contributing to the development of 3D NAND Flash memory technology. Dr. Ray's research interests encompass electronic devices and systems, with a specific focus on enhancing the security, reliability, non-volatility, and energy efficiency of solid-state storage systems. He holds 20 U.S. issued patents and has authored over 80 research papers published in international journals and conferences. Dr. Ray is the recipient of the NSF CAREER Award (2022), the Best Poster Award at IEEE PAINE (2022) and NSF EPSCoR Research Fellowship (2020). |
|
|
|
Computer Science Department Colloquium Machine Learning in Reverse Engineering Speaker: Jordan Wiens, Founder, Vector 35 Inc When: 11:00AM ~ 11:50AM, Monday February 5, 2024 Abstract: What's working and what isn't to improve the life of reverse engineers from the rapidly changing domain of machine learning. Vector 35 (the smarter parts of the company anyway) has spent the last five years working on applying ML to improve the workflows of reverse engineers. Recent improvements in LLMs from OpenAI and other similar models necessitated throwing out some (but not all!) of the techniques, but there's more to just a good LLM in terms of solving real world problems. This talk will present a number of new features in Sidekick (a Binary Ninja plugin), what models and techniques are being leveraged, and what problems still exist (there are many!). Bio: Jordan has over 20 years of experience in the field of information security, spanning network security, forensics, incident response, penetration testing, reverse engineering and vulnerability research. After playing both offense and defense, he has spent the last 8 years building Binary Ninja, a decompiler used for both offensive and defensive applications in information security. Jordan is NOT an expert in machine learning, but he's been active in reverse engineering and vulnerability research for a long time with private research predating pwnie-award winning results and multiple DEF CON black badges from CTF. |
|
|
|
Computer Science Department Colloquium Sustaining Software Teams and Engaging a Diverse Set of People Speaker: Bianca Trinkenreich, Postdoctoral Scholar at Oregon State University When: 9:00AM ~ 10:00AM, Monday February 12, 2024 Abstract: Software development teams represent socio-technical ecosystems that need a complex balance between team engagement and turnover management to ensure a sustainable and productive workforce. High team attrition costs include workflow disruption, knowledge loss, and involve recruitment and onboarding expenses. Research and practice show that the interaction between people matters. Unhappy team members contaminate others, resulting in low productivity and low software quality. During the talk, I will explore the intricate, nuanced, and dynamic factors shaping decisions to participate or disengage within software development teams. Motivations, perceptions of success, belongingness, encountered challenges, and reasons for disengagement across different demographics underscore the impracticality of adopting a 'one size fits all' approach for fostering a sustainable team. Bio: Bianca is a Postdoctoral Scholar at Oregon State University. Her main research interest lies in the Human Aspects of Software Engineering, with the goal of understanding sustainability in software engineering teams, leveraging Diversity and Inclusion, Technostress, Developer Productivity, and Developer Experience. Bianca is the first author of papers published in prestigious venues in Software Engineering, and her contributions have been acknowledged with awards, including the ACM SIGSOFT Outstanding Doctoral Dissertation Award 2024, the Distinguished Paper Award at ICSE Technical Track 2023, the Best Paper Award at ICSE SEIS 2022, and an Honorable Mention Award at CSCW 2020. With 20 years of experience in the IT industry before her research career, Bianca possesses a unique ability to bridge the gap between academia and industry. Her passion for interdisciplinary research and collaboration with the industry enables her to address practical needs and transform them into valuable research opportunities. |
|
|
|
ISTeC Distinguished Lecture, and Electrical and Computer Engineering Department and Computer Science Department Colloquium The Hitchhiker’s Guide to the Multiverse: From Gutenberg’s Printing Press to Metaverse’s Peak-Experience Machine Speaker: Martin Maier, Professor, Institut National de la Recherche Scientifique (INRS), Montréal, Canada. When: 11:00AM ~ 11:50AM, Monday February 12, 2024 Abstract: The twenty-first century has been hailed by renowned historians like Niall Ferguson as the “Age of Networks.” Our era is the “Second Networked Era,” with the personal computer in the role of the printing press and a world connected as never before through the Internet. We don’t yet know what the Internet truly is. While Gutenberg’s printing press gave birth to printing, the Internet’s full potential still remains to be unleashed in the years to come. Measured in Gutenberg time, we stand today at about the year 1483 with the progression of disruption in society. Note that Luther was born in the year 1483. Hence, the Internet’s Martin Luther is yet to come. This Distinguished Lecture starts with Johannes Gutenberg’s printing press, which played a pivotal role in the “First Networked Era” not only in the spread of knowledge but also in Martin Luther’s reformation of society, heralding 300 years of Renaissance. We then briefly review the history of past media revolutions and show that the next Internet revolution will be the so-called Web3 Token Economy, the successor of today’s Web2 Platform Economy and the harbinger of the emerging Metaverse, the predecessor of tomorrow’s Multiverse. We present our ideas and findings on how to realize the reality-virtuality continuum of the Metaverse, the creation of tokenized digital twins for raising our collective intelligence, as well as the beneficial impact of persuasive social robots and advanced extended reality (XR) wearables on the pro-social behavior and cognitive assistance of humans. We conclude by sharing our thoughts on how 6G and Next G networks’ audacious goals aim at delivering digital world experiences (DWEs) across physical, digital, and biological worlds that are at once brand new and very ancient, thereby giving access to the upper range of human peak-experiences that may lift the veil of Metareality and may place us in the midst of the next Renaissance referred to as Eleusis 2.0. Bio: Martin Maier is a full professor with the Institut National de la Recherche Scientifique (INRS), Montréal, Canada. He was educated at the Technical University of Berlin, Germany, and received MSc and PhD degrees both with distinctions (summa cum laude) in 1998 and 2003, respectively. In 2003, he was a postdoc fellow at the Massachusetts Institute of Technology (MIT), Cambridge, MA. He was a visiting professor at Stanford University, Stanford, CA, 2006 through 2007. He was a co-recipient of the 2009 IEEE Communications Society Best Tutorial Paper Award. Further, he was a Marie Curie IIF Fellow of the European Commission from 2014 through 2015. In 2017, he received the Friedrich Wilhelm Bessel Research Award from the Alexander von Humboldt (AvH) Foundation in recognition of his accomplishments in research on FiWi-enhanced mobile networks. In 2017, he was named one of the three most promising scientists in the category “Contribution to a better society” of the Marie Skłodowska-Curie Actions (MSCA) 2017 Prize Award of the European Commission. In 2019 ⁄ 2020, he held a UC3M-Banco de Santander Excellence Chair at Universidad Carlos III de Madrid (UC3M), Madrid, Spain. Recently, he was awarded with the 2023 Technical Achievement Award of the IEEE Communication Society Tactile Internet Technical Committee for his contributions to 6G ⁄ Next G and the Design of Metaverse Concepts and Architectures as well as the 2023 Outstanding Paper Award of the IEEE Computer Society Bio-Inspired Computing Special Technical Community for his contributions to the Symbiosis of INTERnet and human BEING (INTERBEING). He is co-author of the book “Toward 6G: A New Era of Convergence” (Wiley-IEEE Press, January 2021) and author of the sequel “6G and Onward to Next G: The Road to the Multiverse” (Wiley-IEEE Press, February 2023). |
|
|
|
Electrical and Computer Engineering Department and Computer Science Department Colloquium Remembering the Future: From Man-Computer Symbiosis in the 60’s to Symbiotic INTERBEING in 6G Speaker: Martin Maier, Professor, Institut National de la Recherche Scientifique (INRS), Montréal, Canada. When: 9:30-10:30 am, Tuesday February 13, 2024 Abstract: Increasingly, conversations about AI take on the tone and themes of Elon Musk-like predictions about a future in which intelligent robots replace humans first as workers, and then as a species. What is less known is that this Turing-derived vision of AI is not an inevitable future, but one option of two competing visions of AI. For our species’ sake, we ought to remember the future envisioned by two AI pioneers in the 60’s and before. While Turing’s vision of an anthropomorphic (human-equivalent) automative AI puts an emphasis on automation and imitation of existing human behavior, Licklider’s alternative vision of AI puts an emphasis on augmentation and discovery of new human behavior, aiming at a hybrid (human-machine) creative AI that collaborates in symbiosis with humans to usher in a post-human future of cyborg (cybernetic organism) intelligence instead of Turing’s android (machine) superintelligence. In this seminar, we first review our work on the so-called human-in-the-loop Tactile Internet, which adds a new dimension to human-to-machine interaction by enabling tactile and haptic sensations, for a race with (rather than against) machines. Further, we explore the symbiosis of blockchain technologies, most notably the decentralized autonomous organization (DAO), with other key technologies such as AI and robots, putting our focus on the Tactile Internet for advanced human-to-machine interaction via crowdsourcing of human expertise and enhancing the human capabilities of unskilled crowd members. Finally, we present our concept of INTERBEING based on the symbiosis between INTERnet and human BEING, which leverages on hyperintelligent life-like cybernetic organisms to lay the foundation for the intelligentization of future 6G networks by benefitting from not only recently emerging generative AI (high-tech) but also nature’s more-than-human intelligence derived from its sacred way of knowing shaped by eons of evolution (no-tech). Bio: Martin Maier is a full professor with the Institut National de la Recherche Scientifique (INRS), Montréal, Canada. He was educated at the Technical University of Berlin, Germany, and received MSc and PhD degrees both with distinctions (summa cum laude) in 1998 and 2003, respectively. In 2003, he was a postdoc fellow at the Massachusetts Institute of Technology (MIT), Cambridge, MA. He was a visiting professor at Stanford University, Stanford, CA, 2006 through 2007. He was a co-recipient of the 2009 IEEE Communications Society Best Tutorial Paper Award. Further, he was a Marie Curie IIF Fellow of the European Commission from 2014 through 2015. In 2017, he received the Friedrich Wilhelm Bessel Research Award from the Alexander von Humboldt (AvH) Foundation in recognition of his accomplishments in research on FiWi-enhanced mobile networks. In 2017, he was named one of the three most promising scientists in the category “Contribution to a better society” of the Marie Skłodowska-Curie Actions (MSCA) 2017 Prize Award of the European Commission. In 2019 ⁄ 2020, he held a UC3M-Banco de Santander Excellence Chair at Universidad Carlos III de Madrid (UC3M), Madrid, Spain. Recently, he was awarded with the 2023 Technical Achievement Award of the IEEE Communication Society Tactile Internet Technical Committee for his contributions to 6G ⁄ Next G and the Design of Metaverse Concepts and Architectures as well as the 2023 Outstanding Paper Award of the IEEE Computer Society Bio-Inspired Computing Special Technical Community for his contributions to the Symbiosis of INTERnet and human BEING (INTERBEING). He is co-author of the book “Toward 6G: A New Era of Convergence” (Wiley-IEEE Press, January 2021) and author of the sequel “6G and Onward to Next G: The Road to the Multiverse” (Wiley-IEEE Press, February 2023). |
|
|
|
Computer Science Department Colloquium Evolution and Prospects of Large Language Models in Software Engineering Speaker: Toufique Ahmed, Post-doctoral Scholar, University of California, Davis When: 10:00AM ~ 11:00AM, Tuesday February 13, 2024 Abstract: Large Language Models (LLMs) have revolutionized the field of Natural Language Processing (NLP) in recent years. These models are increasingly prominent in software development tasks, including code completion, program repair, code summarization, and incident management. We have witnessed significant changes and the introduction of innovative concepts such as few-shot learning, chain-of-thought, automatic prompt augmentation (ASAP), and multilingual training. In this presentation, I will provide a brief overview of these concepts and how LLMs have evolved in the context of comment generation tasks. I will also explore future possibilities in this area. Additionally, I will delve into the practical applications of LLMs, particularly in critical scenarios like incident management, and gain insights from incident owners regarding their perspectives on this technology. Bio: Toufique Ahmed is a postdoctoral scholar and a former PhD student at University of California, Davis advised by Prof. Prem Devanbu. He is primarily interested in applying and understanding Large Language Models (LLMs) in the context of Software Engineering. He was the recipient of the five-year prestigious Dean’s Distinguished Graduate Fellowship (DDGF) offered by The College of Engineering, UC Davis. He worked with Tom Zimmermann as a research intern at Microsoft Research in 2022. He received his B.Sc. and M.Sc. in Computer Science and Engineering from Bangladesh University of Engineering and Technology (BUET) in 2014 and 2016. His work has been published in the most prestigious SE conferences, including ICSE, ESEC ⁄ FSE, ASE, as well as journals such as TSE and EMSE. He was granted the Graduate Group of Computer Science (GGCS) Outstanding Graduate Research Award (honorable mention) from UC Davis in 2022. He is actively collaborating with Microsoft Research, University College London, Columbia University, and University of Houston. |
|
|
|
Computer Science Department Colloquium Analyzing Safety of Learning-enabled Systems: Bottom-up and Top-down Perspectives Speaker: Ravi Mangal, Postdoctoral Researcher, Carnegie Mellon University When: 10:00AM ~ 11:00AM, Friday February 16, 2024 Abstract: As deep neural networks (DNNs) demonstrate growing capabilities to solve complex tasks, there is a push to incorporate them as components in software and cyber-physical systems. To reap the benefits of these learning-enabled systems without propagating harms, there is an urgent need to develop tools and methodologies for evaluating their safety. Formal methods are a powerful set of tools for analyzing behaviors of software systems. However, formal analysis of learning-enabled systems is challenging; DNNs are notoriously difficult to interpret and lack logical specifications, the environments in which these systems operate can be difficult to model mathematically, and existing formal methods do not scale to these complex systems. In this talk, I will present a bottom-up and a top-down perspective for the analysis of such systems. The bottom-up perspective focuses on analyzing DNNs in isolation. To address the challenges in intepreting and specifying DNN behavior, I will present a logical specification language designed to facilitiate writing specifications about vision-based DNNs in terms of high-level, human-understandable concepts. I will then demonstrate how we can leverage vision-language models such as CLIP to encode and check these specifications. The top-down perspective focuses on analyzing learning-enabled systems as a whole. To address the challenges in modeling the environment and scaling formal analysis, I will present new probabilistic abstractions for DNN-based perception components in learning-enabled cyber-physical systems that make it feasible to formally analyze such systems. Bio: Ravi Mangal is a postdoctoral researcher at Carnegie Mellon University in the Security and Privacy Institute (CyLab) hosted by Dr. Corina Pasareanu. He graduated with a PhD in Computer Science from Georgia Institute of Technology in 2020 advised by Dr. Alex Orso. He is interested in all aspects of designing and applying formal methods for assuring the correctness and safety of software systems. His current research focuses on developing algorithms and methodologies for formally analyzing the safety and trustworthiness of learning-enabled systems. |
|
|
|
Computer Science Department Colloquium Towards Practical Software Quality Assurance Techniques via Automated Support of the Development and Usage Process Speaker: Austin Mordahl, PhD Candidate, University of Texas at Dallas When: 9:00AM ~ 10:00AM, Monday February 19, 2024 Abstract: Automated software quality assurance approaches are an important guardrail against the proliferation of bugs and security issues in software. However, challenges like unreliable results, difficulty in configuration, and high false positive rates prevent many of these techniques from seeing widespread adoption. In this talk, I will present my work aimed at breaking these barriers to the usage of automated software quality assurance techniques, with a specific focus on static analysis. My work improves automated software quality assurance techniques along various dimensions throughout their development and usage processes; specifically, with regard to their reliability, usability, and applicability. First, I will share my experience improving the reliability of static analysis tools through the development of an automated testing and debugging framework, enabled by a novel theoretical model of static analysis tools’ configuration spaces. This framework has allowed the detection of dozens of bugs in popular static analysis tools. Second, I will detail the work that aims to improve usability by adapting machine learning models to automate tedious, manual tasks in the usage process of static analysis tools, including configuring these tools for specific target programs and classifying false positives. Finally, I will provide my vision for the future of automatic software quality assurance techniques, which involves the improvement of the development and usage processes through automated, adaptive, and explainable techniques that provide a seamless developer and user experience. I will touch on three concrete research projects I plan to undertake in the next 5 years towards this vision. Bio: Austin Mordahl is a final-year Ph.D. candidate at the University of Texas at Dallas. His research is in the area of Software Engineering. Specifically, he focuses on advancing the state-of-the-art in automated software quality assurance techniques (e.g., static program analysis and fuzz testing) through various techniques such as machine learning, and software testing and debugging. He regularly publishes at the top-tier software engineering conferences and journals, such as ICSE, FSE, ASE, ISSTA, and EMSE. He was a recipient of the prestigious National Science Foundation (NSF) Graduate Research Fellowship, and the Eugene McDermott Graduate Research Fellowship in 2020. Moreover, he won the ACM Student Research Competition at ICSE 2019. More information is available at https: ⁄ ⁄ austinmordahl.com. |
|
|
|
ISTeC Distinguished Lecture, and Computer Science Department and Electrical and Computer Engineering Department Colloquium Advancing Machine Learning in the End-of-Moore Era: Challenges and Opportunities Speaker: P. Sadayappan, Professor, Kahlert School of Computing, University of Utah When: 11:00AM ~ 11:50AM, Monday February 19, 2024 Abstract: Although artificial neural networks were invented over fifty years ago, it was not until computers were sufficiently powerful that the deep learning revolution could get started earlier in this century. Thanks to steady increase in computational power over the last 15 years, increasingly complex and powerful deep learning models have had a transformative impact on virtually all aspects of society. However, it is getting more and more difficult to maintain the dramatic rates of increase in compute power made possible by smaller transistor sizes with successive generations of VLSI technology, because we are approaching transistor feature sizes close to atomic limits. Therefore other avenues will become more critical in enabling more powerful machine learning models in the future. In this talk, we will discuss approaches to make more effective use of bounded hardware resources, including exploitation of sparsity, algorithm-architecture co-design, and the use of machine learning to improve the performance of machine learning applications. Bio: Sadayappan is a Professor in the Kahlert School of Computing at the University of Utah, with a joint appointment at Pacific Northwest National Laboratory. His primary research interests center around compiler optimization for high-performance computing, with an emphasis on performance optimization of matrix ⁄ tensor computations arising in applications from scientific computing and machine learning. He is the recipient of an ACM SIGPLAN Most Influential PLDI Paper award. Sadayappan is an IEEE Fellow. |
|
|
|
Computer Science Department and Electrical and Computer Engineering Department Colloquium sponsored by ISTeC Compiler Optimization of Tensor Computations Speaker: P. Sadayappan, Professor, Kahlert School of Computing, University of Utah When: 9:30AM ~ 11:00 AM, Tuesday February 20, 2024 Abstract: Production compilers like clang and gcc are extremely effective in generating very compact machine code from high-level C ⁄ C++ programs, i.e., they are very effective in minimizing the number of executed instructions. However, the dominant cost (both in terms of energy and execution time) on all computer systems today is not that of execution of the needed arithmetic operations but of the movement of data, between processors of a parallel system and through the memory hierarchy at each processor. Despite significant research advances in compiler optimization for affine computations, such as the powerful polyhedral model for dependence analysis and loop transformation, it remains extremely challenging for any compiler today to generate optimized code (for either multicore CPUs or GPUs) that achieves performance comparable to manually architected vendor libraries or autotuning optimizers like TVM. In this talk, we will discuss challenges and opportunities for compiler optimization of tensor computations, including design-space exploration, performance modeling, and algorithm-architecture co-design. Bio: Sadayappan is a Professor in the Kahlert School of Computing at the University of Utah, with a joint appointment at Pacific Northwest National Laboratory. His primary research interests center around compiler optimization for high-performance computing, with an emphasis on performance optimization of matrix ⁄ tensor computations arising in applications from scientific computing and machine learning. He is the recipient of an ACM SIGPLAN Most Influential PLDI Paper award. Sadayappan is an IEEE Fellow. |
|
|
|
Computer Science Department Colloquium Towards Multimodal Interaction for Augmented and Virtual Reality and Beyond Speaker: Francisco Ortega, Assistant Professor, Department of Computer Science, Colorado State University When: 11:00AM ~ 11:50AM, Monday February 26, 2024 Abstract: Dr. Ortega’s presentation will cover some of the research areas he has conducted over 5 ½ years at Colorado State University. The talk will describe completed and ongoing efforts and directions for the next 5-10 years. One of Dr. Ortega’s primary research areas focuses on multimodal interaction (gesture-centric), a critical component in his view of realizing everyday augmented reality (AR). While controllers remain the most common way to interact in virtual reality (VR), AR (and some VR) head-mounted displays (HMDs) include mid-air gesture interactions where the user extends their arm to interact with the virtual scene. Some systems include basic multimodal interaction (e.g., gaze + gesture pinch). However, rich multimodal interaction in AR ⁄ VR still requires additional studies to deliver intuitive 3D user interfaces for virtual environments and everyday applications. The design of multimodal and unimodal interaction is key to developing new experiences and expanding access to the "Invisible Computer" – a long-term dream of the late Dr. Weiser. In addition to Dr. Ortega’s multimodal interaction research, he will briefly describe some highlights of the other areas he currently pursues, including notifications and VR forest bathing. Dr. Ortega will also cover some lessons learned that may be useful to other researchers. Bio: Francisco R. Ortega is an Assistant Professor at Colorado State University (CSU) and has been Director of the Natural User Interaction lab (NUILAB) since Fall 2018. Dr. Ortega earned his Ph.D. in Computer Science (CS) in the field of Human-Computer Interaction (HCI) and 3D User Interfaces (3DUI) from Florida International University (FIU). He also held the Post-Doc and Visiting Assistant Professor position at FIU between February 2015 and July 2018. His research has focused on (1) multimodal and unimodal interaction (gesture-centric), which includes gesture elicitation (e.g., a form of participatory design), (2) information access effort in augmented reality (e.g., visual cues and automation bias), (3) AR notifications, and (4) stress reduction using virtual reality forest bathing. For multimodal interaction research, Dr. Ortega focuses on improving user interaction by (a) multimodal elicitation, (b) developing interactive techniques, and (c) improving augmented reality visualization techniques. The primary domains for interaction include general environments, immersive analytics, and VR sketching. His research has resulted in over 90 peer-reviewed publications, including books, journals, conferences, workshops, and magazine articles, in venues such as ACM CHI, ACM VRST, IEEE VR, IEEE TVCG, IEEE ISMAR, ACM PACMHCI, ACM ISS, ACM SUI, IEEE 3DUI, HFES, and Human Factor Journals, among others. Dr. Ortega has experience with multiple projects awarded by the government. For example, Dr. Ortega was a co-PI for the DARPA Communicating with Computers project. He is a PI for a 3-year effort for ONR titled Perceptual ⁄ Cognitive Aspects of Augmented Reality: Experimental Research and a Computational Model. He was recently awarded a new ONR grant titled “Assessing Cognitive Load and Managing Extraneous Load to Optimize Training.” The National Science Foundation and other agencies and companies have also funded him. This includes the NSF CAREER 2023 for microgestures and multimodal interaction. Since his tenure-track appointment at CSU in August 2018, Dr. Ortega has brought over 4.2 million dollars in external funding (with 3.7 million as principal investigator). Finally, Dr. Ortega is committed to diversity and inclusion, and his mission is to increase the number of underrepresented minorities in CS, rooted in his own experiences and from the time spent at FIU – the largest R1 Hispanic serving institution. |
|
|
|
Computer Science Department Colloquium Supporting the Task-driven Skill Identification in Open Source Project Issue Tracking Systems Speaker: Fabio Santos, Machine Learning Engineer and Instructor at Grand Canyon University When: 10:00AM ~ 11:00AM, Monday March 4, 2024 Abstract: Selecting an appropriate task is challenging for contributors to Open Source Software (OSS), mainly for those contributing for the first time. Therefore, researchers and OSS projects have proposed various strategies to aid newcomers, including labeling tasks. In this research, we investigate the automatic labeling of open issues strategy to help the contributors pick a task to contribute. We label the issues with API domains---categories of APIs parsed from the source code used to solve the issues. We employed mixed methods. We qualitatively analyzed interview transcripts and the survey's open-ended questions to comprehend the strategies communities use to assist in onboarding contributors. We applied quantitative studies to analyze the relevance of the API-domain labels in a user experiment and compared the strategies' relative importance for diverse contributor roles. We mined project and issue data from OSS repositories to build the ground truth and predictors to infer the API-domain labels and compared precision, recall, and F-measure with state-of-the-art. Additionally, inspired by previous research, label prediction might benefit from leveraging metrics derived from communication data and social network analysis (SNA) for issues in which social interaction occurs. Thus, we study how these "social metrics'' can improve the automatic labeling of open issues with API domains. We mined conversation data from OSS projects' repositories and organized it in periods to reflect the contributors' project participation seasonality. We replicated social metrics from previous work and added them to the corpus to predict API-domain labels. Social metrics improved the performance of the classifiers compared to using only the issue description text in terms of precision (0.922), recall (0.978), and F-measure (0.942). These results indicate that social metrics can help capture the patterns of social interactions in a software project and improve the labeling of issues in an issue tracker. Future work includes using a skill ontology to assist the matching process between contributors and tasks encompassing multi-level skills and expertise. By quantitatively analyzing the confidence level of the matching instances in ontologies describing contributors' skills and tasks, we might recommend issues for contribution. In addition, we will measure the effectiveness of the API-domain labels by evaluating contributions and measuring the progress and correctness among the labeled and unlabelled ones. The results showed that assigning labels to issues is an essential strategy for diverse roles in OSS communities. The API-domain labels are relevant, mainly for experienced practitioners. Labeling the issues with the API-domain labels indicates the skills involved in an issue. The labels represent possible libraries (aggregated into domains) used in the source code related to an issue. We also implemented a free, open-source software demonstration tool to assist newcomers in finding a task to start based on a set of skills elected in a user interface. By investigating this research topic, we expect to assist OSS communities in attracting and onboarding new contributors. Bio: Fabio Santos earned his Bachelor's and Master's degrees in Informatics, specializing in databases, from the Pontifícia Universidade Católica do Rio de Janeiro, Brazil. He obtained his Ph.D. in Informatics, focusing on Information Systems, Knowledge Modeling, and Reasoning, from the Universidade Federal do Estado do Rio de Janeiro, Brazil, in 2022. Additionally, he completed a Ph.D. in Computer Science at Northern Arizona University, USA, in 2023. With a career spanning over two decades in the IT sector, Santos started as a Software Engineer and eventually became the IT Superintendent for the Brazilian Navy. His research interests lie in knowledge modeling for system and ontology network integration, applying artificial intelligence in software engineering, open-source software, recommendation systems, mining software repositories, and analyzing social networks. |
|
|
|
Computer Science Department Colloquium Decentralization in Delegated-Proof-of-Stake (DPoS) Blockchains: Myths Vs Reality Speaker: Balaji Palanisamy, Associate Professor, School of Computing and Information, University of Pittsburgh When: 11:00AM ~ 11:50AM, Monday March 4, 2024 Abstract: Coin-based voting governance is the building block of Delegated-Proof-of-Stake (DPoS) blockchains such as TRON, Steem and EOSIO. We recently witnessed the historical event of the first de facto blockchain takeover between Steem and TRON. Within one hour of this incident, TRON founder took over the entire Steem committee, forcing the original Steem community to leave the blockchain that they maintained for years. In this talk, we will discuss our original findings on the study performed on Steemit that revealed that the actual level of decentralization in Steemit is far below the ideal level, exposing the vulnerabilities of the Delegated Proof-of-Stake (DPoS) consensus protocol system years before the actual TRON’s takeover event occurred in Steem. A takeover in DPoS blockchains refers to an attacker controlling the supermajority of block producers and as a result, gaining immense control of the blockchain including the ability to reverse confirmed transactions and change the private keys of accounts. We will present a formal three-phase model for coin-based voting governance and discuss the takeover attack and resistance model using our rigorous analysis of TRON’s takeover of Steem. We will discuss our large-scale empirical study of the passive takeover resistance of EOSIO, Steem and TRON. Finally, we will present our novel insights into the security of coin-based voting governance and discuss potential future work on improving the takeover resistance of DPoS blockchains. Bio: Balaji Palanisamy is an Associate Professor in the School of Computing and Information at the University of Pittsburgh. His research interests include Blockchains, data privacy, privacy-preserving system design and scalable resource management for distributed systems. He obtained his Ph.D. from Georgia Institute of Technology in 2013. At the University of Pittsburgh, he carries out research in the Laboratory of Research and Education on Security Assured Information Systems (LERSAIS). He is a recipient of IBM Faculty Award, and his research has received the Distinguished paper award at CCS 2023, and Best Paper Awards at DBSec 2022, IEEE BigDataCongress 2018, IEEE BigDataCongress 2017, IEEE ⁄ ACM CCGrid 2015 and IEEE CLOUD 2012. He is an Associate Editor for the IEEE Transactions on Dependable and Secure Computing, IEEE TDSC and the IEEE Transactions on Services Computing, IEEE TSC journals. |
|
|
|
Computer Science Department Colloquium Cognitive Engineering at NRL: Research Overview Speaker: Dina Acklin and Rebecca R. Goldstein When: 11:00AM ~ 11:50AM, Thursday March 7, 2024 Abstract: Cognitive engineers from the US Naval Research Lab will discuss an overview of the ongoing research being conducted at NRL-Stennis within the Cognitive and Geospatial Sciences area. A summary of relevant projects will be provided, with a focus on current collaborative work with CSU in the area of developing models of human visual saliency to be integrated into APIs for text overlays in 2D, VR, and AR environments. Bio: Dina Acklin is an Engineering Psychologist and Human Factors researcher at the Naval Research Laboratory at Stennis Space Center, MS. As a deputy program officer at the Office of Naval Research, she has worked to shape research programs in the areas of command decision making, human-machine teaming, collaborative AI, and applied machine learning. Her current projects involve applying findings from basic cognitive research to applied problems facing the US Navy, Air Force, and Marine Corp. Her current research involves investigating the perception of various sound classes for the development of auditory camouflage, the application of cognitive feedback to machine learning models for geospatial applications, and developing strategies for displaying features in AR and VR environments driven by human visual perception capabilities. Rebecca R. Goldstein is Engineering Research Psychologist with a PhD in Cognitive Psychology from Louisiana State University. She has an extensive background utilizing eye tracking to address whether eye tracking metrics can distinguish between visual search targets stored in working memory or from long term memory (Goldstein & Beck, 2018), to compare the performance between experts and novices searching through maps that varied in clutter (Beck et al., 2012), and to assess whether participants would adopt a global or local attention allocation strategy to complete a task given no feedback, feedback, or a reward (Beck, Goldstein, et al., 2018). Her current research involves bridging the gap between basic cognitive research and applied problems facing the US Navy, and Marine Corp. She currently contributes to projects addressing strategies for displaying features in VR and AR environments, understanding human categorization and identification of auditory sounds, consideration of explicit and implicit human feedback on machine learning models for geospatial applications, and investigating human performance with visual and auditory demanding tasks. |
|
|
|
Computer Science Department Colloquium CANCELED Speaker: CANCELED When: 11:00AM ~ 11:50AM, Monday March 18, 2024 Abstract: |
|
|
|
ISTeC Distinguished Lecture, and Electrical and Computer Engineering Department and Computer Science Department Colloquium Spectrum challenges for the next generation of wireless systems: sharing between commercial, federal and scientific users Speaker: Monisha Ghosh, Professor of Electrical Engineering, University of Notre Dame When: 11:00AM ~ 11:50AM, Monday March 25, 2024 Abstract: The electromagnetic spectrum, from about 10 kHz to 1 THz, is used for a variety of services, of which commercial terrestrial wireless systems, primarily cellular and Wi-Fi, occupy only a small portion. Other uses of spectrum are less visible to consumers, but equally important, for example, scientific uses like radioastronomy and weather forecasting, and federal uses for safety and life-critical applications, including GPS. Each one of these services have increasing spectrum needs, which are becoming more difficult to manage using the traditional approaches of spectrum allocation, leading to spectrum conflicts such as the recent one between 5G and aircraft altimeters. This talk will provide an overview of the spectrum challenges faced by both regulators and technologists as the next generation of systems are being developed, along with potential solutions that harness technological advancements that permit spectrum sharing between different use-cases. Bio: Monisha Ghosh is a Professor of Electrical Engineering at the University of Notre Dame and a member of the Notre Dame Wireless Institute. She is also the Policy Outreach Director for SpectrumX, the first NSF Center for Spectrum Innovation and the co-chair of the FCC’s Technological Advisory Council (TAC) Working Group on Advanced Spectrum Sharing. Her research interests are in the development of next generation wireless systems: cellular, Wi-Fi and IoT, with an emphasis on spectrum sharing and coexistence and applications of machine learning to improve network performance. Prior to joining the University of Notre Dame in 2022, she was the Chief Technology Officer at the Federal Communications Commission, a Program Director at the National Science Foundation, Research Professor at the University of Chicago and spent 24 years in industry research at Bell Labs, Philips Research and Interdigital working on a wide variety of wireless systems: HDTV, Wi-Fi, TV White Spaces and cellular. She obtained her B.Tech from IIT Kharagpur in 1986 and Ph.D. from USC in 1991. She is a Fellow of the IEEE. |
|
|
|
Electrical and Computer Engineering Department and Computer Science Department Colloquium Real-world performance of 4G and 5G in mmWave, mid-band and shared spectrum (CBRS) Speaker: Monisha Ghosh, Professor of Electrical Engineering, University of Notre Dame When: 9:30-10:30 am, Tuesday March 26, 2024 Abstract: As 5G deployments increase in the newly allocated mid-band and mmWave spectrum, and discussions on 6G begin, it is important to characterize real-world performance of the enhancements made to 5G to determine how best to design the next generation of cellular networks. For example, how well does massive MIMO work in the real-world? How is sharing in CBRS performing? In this talk we will present recent results from detailed measurements of 4G and 5G in the various bands: mmWave (< 24 GHz), mid-band (2.5 - 3.98 GHz) and CBRS (3.55 - 3.7 GHz). Our studies demonstrate that mmWave 5G is severely limited in coverage, especially indoors, while performance of 5G in mid-band also depends on network densification, contrary to popularly held beliefs. Our studies in CBRS show that secondary co-channel sharing as well as adjacent channel interference from high power deployments pose fundamental challenges for cellular networks in shared spectrum. We conclude with some directions for future network design based on our research that will allow 6G to be "sharing native". Bio: Monisha Ghosh is a Professor of Electrical Engineering at the University of Notre Dame and a member of the Notre Dame Wireless Institute. She is also the Policy Outreach Director for SpectrumX, the first NSF Center for Spectrum Innovation and the co-chair of the FCC’s Technological Advisory Council (TAC) Working Group on Advanced Spectrum Sharing. Her research interests are in the development of next generation wireless systems: cellular, Wi-Fi and IoT, with an emphasis on spectrum sharing and coexistence and applications of machine learning to improve network performance. Prior to joining the University of Notre Dame in 2022, she was the Chief Technology Officer at the Federal Communications Commission, a Program Director at the National Science Foundation, Research Professor at the University of Chicago and spent 24 years in industry research at Bell Labs, Philips Research and Interdigital working on a wide variety of wireless systems: HDTV, Wi-Fi, TV White Spaces and cellular. She obtained her B.Tech from IIT Kharagpur in 1986 and Ph.D. from USC in 1991. She is a Fellow of the IEEE. |
|
|
|
Computer Science Department Colloquium Covariate Software Vulnerability Discovery Model to Support Cybersecurity Test & Evaluation Speaker: Lance Fiondella, Associate Professor, Department of Electrical & Computer Engineering, University of Massachusetts Dartmouth. Founding Director, University of Massachusetts Dartmouth Cybersecurity Center. When: 10:00AM ~ 10:50AM, Thursday March 28, 2024 Abstract: Vulnerability discovery models (VDM) have been proposed as an application of software reliability growth models (SRGM) to software security-related defects. VDM model the number of vulnerabilities discovered as a function of testing time, enabling quantitative measures of security. Despite their obvious utility, past VDM have been limited to parametric forms that do not consider the multiple activities software testers undertake in order to identify vulnerabilities. In contrast, covariate SRGM characterize the software defect discovery process in terms of one or more test activities. However, data sets documenting multiple security testing activities suitable for the application of covariate models are not readily available in the open literature. To demonstrate the applicability of covariate SRGM to vulnerability discovery, this research identified a web application to target as well as multiple tools and techniques to test for vulnerabilities. The time dedicated to each test activity and the corresponding number of unique vulnerabilities discovered were documented and prepared in a format suitable for the application of covariate SRGM. Analysis and prediction were then performed and compared with a flexible VDM without covariates, namely the Alhazmi-Malaiya Logistic Model (AML). Our results indicate that covariate VDM significantly outperformed the AML model on predictive and information-theoretic measures of goodness of fit, suggesting that covariate VDM are a suitable and effective method to predict the impact of applying specific vulnerability discovery tools and techniques. Bio: Lance Fiondella is an associate professor in the Department of Electrical & Computer Engineering at the University of Massachusetts Dartmouth and the Founding Director of the University of Massachusetts Dartmouth Cybersecurity Center, A NSA ⁄ DHS designated Center of Academic Excellence in Cyber Research (CAE-R). He received his PhD (2012) in Computer Science & Engineering from the University of Connecticut. Dr. Fiondella has published over 160 peer-reviewed journal articles and conference papers, fourteen of which have been recognized with awards, including five as first author and eight with a major advisee as first author. His research has been funded by DHS, ARL, USMA, ERDC, NAVAIR, NAVSEA, Air Force, NASA, and NSF, including a CAREER award as well as a $3.5 million dollar NSF CyberCorps(R) Scholarship for Service (SFS) project to recruit and train the next generation of cybersecurity professionals. Dr. Fiondella has also held various appointments with U.S. Government Laboratories and Federally Funded Research & Development Centers. |
|
|
|
Computer Science Department Colloquium Programming the Built Environment with Knowledge Graphs Speaker: Gabe Fierro, Assistant Professor of Computer Science, Colorado School of Mines. Joint appointment with NREL. When: 11:00AM ~ 11:50AM, Monday April 1, 2024 Abstract: The built environment has a data problem. The buildings, cities, water treatment plants, and other human-made systems produce more data now than ever before, opening new possibilities of using data to optimize operation, reduce energy consumption, predict performance, and identify faults. However, the complexity, heterogeneity, and high degree of churn of these systems makes it expensive and difficult to develop software for them. Models, control sequences, data analytics, and other software-based solutions must often be rewritten from scratch for each environment in which they will be deployed. The process of discovering and accessing data is further exacerbated by the lack of standardized structured representations of built environment systems. These challenges significantly impede the adoption of data-driven sustainable practices at societal scale. This talk will explore the use of semantic knowledge graphs to normalize descriptions of the built environment, specifically smart buildings, and reduce the cost of developing and deploying data-driven software in these settings. First, I will describe how ontologies can constrain knowledge graphs to produce useful abstractions of complex cyber-physical systems, as typified by the Brick ontology for smart buildings. Elements of this work are being adapted into new knowledge graph standards for buildings. Next, I will show how knowledge graphs enable novel programming models for "portable software" where programs can adapt their own operation to individual environments, based on queries against the knowledge graph. The talk will also show how these emerging use cases for knowledge graphs contrast with prevailing approaches towards knowledge graph maintenance and management and give rise to new methods for specifying and repairing knowledge graphs. Finally, I will show how these new technologies enable novel applications for smart buildings. Bio: Dr. Gabe Fierro is an assistant professor of Computer Science at Colorado School of Mines and enjoys a joint appointment at the National Renewable Energy Laboratory His research focuses on the design and implementation of efficient systems for data and metadata management, primarily for cyber-physical systems and the Internet of Things. A major element of his research investigates knowledge graph design for enabling data-driven science in cyber-physical systems, and practical systems for managing and accessing knowledge graphs in these contexts. In addition, Dr. Fierro is a founder and the lead maintainer of Brick (https: ⁄ ⁄ brickschema.org), an open-source ontology for smart buildings enabling intelligent and portable data-driven applications. He is also involved in knowledge graph standardization efforts in the smart building and water treatment industries. He received his PhD in Computer Science from UC Berkeley in 2021. |
|
|
|
Computer Science Department Colloquium Learning Switched Models from Data Speaker: Sriram Sankaranarayanan, Professor of Computer Science, Associate Dean for Digital Education, University of Colorado Boulder When: 11:00AM ~ 11:50AM, Monday April 8, 2024 Abstract: The ordinary least squares regression problem seeks a single “straight line” that minimizes the sum of squared errors between given data points and the explaining model. This is fundamental to numerous applications in science and engineering. In this talk, we consider the problem of fitting k < 1 models such that the sum of errors from each point to the “closest” model (best explanation) is minimized. Such problems arise naturally for Cyber-Physical Systems wherein the combination of continuous and discrete behaviors is best explained through multiple behavioral modes. However, this problem known as k-linear regression is NP-hard even for k = 2. We show guaranteed approximation schemes that work in linear time in the number of data points with an exponential dependence on the dimensionality. This is achieved through a combination of cutting planes and ellipsoidal methods. The resulting algorithm also turns out to be practical for relatively large datasets, but in low dimensions. We also look at some of the interesting applications of our work and compare its performance against standard machine learning approaches. The talk will be self-contained, and all relevant concepts will be introduced during the talk. Joint work, mainly with Monal Narasimhamurthy and Guillaume Berger. Bio: Sriram Sankaranarayanan is a professor in the computer science and the associate dean for digital education at the College of Engineering at CU Boulder. His research focusses on the foundations of autonomous systems with an emphasis on applying ideas from Logic and theoretical computer science to design and reason about safety critical systems. He received his PhD in Computer Science from Stanford University in 2005 and subsequently worked at NEC Research Labs in Princeton, NJ before joining CU Boulder in 2009. He has received awards including the NSF CAREER award (2010) and the outstanding innovation award from Coursera (2021) for his online specialization on data structures and algorithms. |
|
|
|
Computer Science Department Colloquium The New Reality of Extended Reality: Empirical Evaluation of Interaction in XR Speaker: Robert J. Teather, Associate Professor and Director - School of Information Technology, Carleton University When: 11:00AM ~ 11:50AM, Monday April 15, 2024 Abstract: Extended reality (XR), a catch-all term for virtual reality (VR), mixed reality (MR) and augmented reality (AR) has recently become popular again with the release of low-cost and effective consumer-grade head-mounted displays such as the Meta Quest. The longstanding dream of VR has users interacting with virtual objects as naturally as real ones. In practice, despite technological advances, numerous technical and human factors make this difficult. Modern VR interaction continues to employ naturally-inspired interaction techniques that have changed little since their introduction in the late 80s. Similarly, cybersickness and the lack of tactile feedback when interacting with virtual objects are well-known to limit the effectiveness of VR systems, yet these issues persist today. In this talk, I will discuss my research addressing these three interrelated areas of virtual reality interaction. I will first describe my studies comparing 3D selection interfaces between 3D and desktop systems, and my work in extending a standardized methodology to support fair and direct comparison between these two different modalities. I will then discuss my research group's recent work employing this standardized methodology for evaluating novel 3D selection methods, as well as other projects aimed at enhancing the usability of VR systems through evaluating the effectiveness of cybersickness reduction techniques and novel approaches to VR haptics that employ shape-changing devices and perceptual illusions. I will close by discussing future directions for this work on both improving usability of, and equitable access to, VR technology. Bio: Robert J. Teather is a leading expert in several interrelated areas of human-computer interaction, including interaction techniques and input devices, especially when applied to 3D user interfaces for virtual reality. He holds a PhD and MSc in Computer Science (York University, Canada), as well as a BSc in Computer Science (Brock University, Canada). His PhD work focused on developing standardized methods for the empirical comparison of input devices for 3D interaction – primarily in order to compare mouse and 3D tracker-based input. To this end, Dr. Teather has established himself as an expert in comparing drastically different input devices and interaction techniques for common fundamental interaction tasks in VR (e.g., target selection), across varying system configurations (e.g., display properties such as stereo graphics, or system properties such as latency). His research is supported by Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Foundation for Innovation. He has also served in lead conference organization roles (e.g., general chair, technical program chair) in events including the IEEE Conference on Virtual Reality & 3D User Interfaces, ACM Virtual Reality Software and Technology, and the ACM Symposium on Spatial User Interaction. |
|
|
|
ISTeC Distinguished Lecture in Conjunction with the Deparment of Journalism and Media Communication, Department of Computer Science, and Department of Electrical and Computer Engineering The Psychology of Trustworthy AI: Resolving the Tension between Human and Machine Agency Speaker: Shyam Sundar, James P. Jimirro Professor of Media Effects and Director of the Center for Socially Responsible Artificial Intelligence, Penn State University When: 11:00AM ~ 11:50AM, Monday April 22, 2024 Abstract: This talk will discuss how intelligent machines pose a threat to human agency, by using media-related examples such as fake news and personalized entertainment. It will propose strategies for reconciling the tension between machine and human agency by presenting theory and research about social and psychological aspects of Human-AI Interaction (HAII). It will focus on psychological issues pertaining to user trust in AI and discuss strategies for promoting socially responsible designs of AI interfaces. Bio: S. Shyam Sundar (PhD, Stanford University) is James P. Jimirro Professor of Media Effects and Director of the Center for Socially Responsible Artificial Intelligence (CSRAI; http: ⁄ ⁄ ai.psu.edu) at Penn State University. He is also the founding director of the Media Effects Research Laboratory at Penn State’s College of Communications (http: ⁄ ⁄ bellisario.psu.edu ⁄ people ⁄ individual ⁄ s.-shyam-sundar). Prof. Sundar is a theorist as well as an experimentalist who uses a variety of quantitative and qualitative approaches in his research. His research examines social and psychological effects of interactive media, ranging from websites and social media to virtual assistants and virtual environments. Specifically, his experiments investigate the role played by technological affordances in shaping user experience of mediated communications. Current research pertains to psychological effects of Human-AI interaction in a variety of contexts, ranging from personalization and recommendation to fake news and content moderation. |
|
|
|
Department of Journalism and Media Communication, Department of Electrical and Computer Engineering, and Department of Computer Science Colloquium Role of AI in Communication: New Theory and Research Speaker: S. Shyam Sundar, James P. Jimirro Professor of Media Effects and Director of the Center for Socially Responsible Artificial Intelligence, Penn State University When: 3:00PM ~ 4:00PM, Monday April 22, 2024 Abstract: This talk will discuss different ways to conceptualize the role of artificial intelligence (AI) in communication research, by drawing upon the speaker’s corpus of concepts and theories pertaining to the effects of new media technologies over the last three decades. It will present key concepts and concerns in the study of AI. It will introduce the speaker’s theoretical framework for human-AI interaction (HAII) based on his theory of interactive media effects (TIME), and describe recent studies that apply his HAII-TIME model to the study of content moderation and recommendation systems. Bio: S. Shyam Sundar (PhD, Stanford University) is James P. Jimirro Professor of Media Effects and Director of the Center for Socially Responsible Artificial Intelligence (CSRAI; http: ⁄ ⁄ ai.psu.edu) at Penn State University. He is also the founding director of the Media Effects Research Laboratory at Penn State’s College of Communications (http: ⁄ ⁄ bellisario.psu.edu ⁄ people ⁄ individual ⁄ s.-shyam-sundar). Prof. Sundar is a theorist as well as an experimentalist who uses a variety of quantitative and qualitative approaches in his research. His research examines social and psychological effects of interactive media, ranging from websites and social media to virtual assistants and virtual environments. Specifically, his experiments investigate the role played by technological affordances in shaping user experience of mediated communications. Current research pertains to psychological effects of Human-AI interaction in a variety of contexts, ranging from personalization and recommendation to fake news and content moderation. |
|
|
|
ISTeC Distinguished Lecture; Computer Science Department and Electrical and Computer Engineering Department Colloquium From Centralized Learning to Federated Learning: Opportunities and Challenges Speaker: Ling Liu When: 11:00AM ~ 11:50AM, Monday April 29, 2024 Abstract: Machine learning has blossomed through (centralized) learning over massive data, evidenced by recent advances in self-supervised multi-modal learning and generative AI powered large language models (LLMs). Most of the benchmark datasets are publicly available data sources and can be freely collected to a centralized Cloud repository to train large models, such as ChatGPT, LLaMA. However, for the missions-critical applications in the real world, massive proprietary data are generated 24x7 at the edge of the Internet. Centralized collection of such geographically distributed and proprietary datasets is neither feasible nor realistic w.r.t. resource ⁄ latency demand and data privacy ⁄ confidentiality requirement. In this distinguished lecture, I will illustrate the potential of self-supervised learning and generative AI, and discuss two important technological advancements in Generative AI, which aim to scale the training and the deployment of large models on the edge. First, we will describe and compare a suite of large model reduction techniques for large foundation models and their fine-tuning of downstream learning tasks. Second, we will introduce Federated learning (FL), an emerging distributed learning paradigm, enabling joint training of a large global model by a distributed population of edge clients, while keeping their sensitive data local and only share their local model updates with the FL server(s). I will conclude with an outlook of generative AI and LLMs. Bio: Ling Liu is a full Professor in the School of Computer Science at Georgia Institute of Technology. She directs the research programs in the Distributed Data Intensive Systems Lab (DiSL), examining various aspects of big data-powered artificial intelligence (AI) systems, and machine learning (ML) algorithms and analytics, including performance, availability, privacy, security, and trust. Prof. Liu is an elected IEEE Fellow, a recipient of IEEE Computer Society Technical Achievement Award (2012), and a recipient of the best paper award from numerous top venues, including IEEE ICDCS, WWW, ACM ⁄ IEEE CCGrid, IEEE Cloud, IEEE ICWS. Currently, Prof. Liu is the editor in chief of ACM Transactions on Internet Computing (since 2019) and the chair of IEEE CS Fellow Evaluation Committee (FY2024). Prof. Liu is a frequent keynote speaker in top-tier venues in Big Data systems, AI ⁄ ML systems and applications, Cloud Computing, Services Computing, Privacy, Security and Trust. Her current research is primarily supported by USA National Science Foundation under CISE programs, CISCO and IBM. |
|
|
|
Computer Science Department and Electrical and Computer Engineering Department Colloquium sponsored by ISTeC Security and Privacy in Federated Learning Speaker: Ling Liu When: 2:00PM ~ 2:50PM, Monday April 29, 2024 Abstract: We have witnessed two existing trends of computing: one is the rapid advances in AI technology fueled by recent generative AI and Large Language Models (LLMs), and the other is the new world of device-edge-cloud computing continuum. These two emerging trends are urging the synergistic alliances of AI and cyber-security in both research and development of next generation of AI-powered device-edge-cloud computing systems. In this talk, I will first discuss privacy and security vulnerabilities in federated learning. Then I will describe the state of the art (SOTA) trustworthy AI methods and techniques against data and model trojan attacks and privacy leakage risks, including lessons learned from our trustworthy AI research projects. I will conclude with an outlook of security and privacy challenges in the rapid growth of generative AI and LLMs. Bio: Ling Liu is a full Professor in the School of Computer Science at Georgia Institute of Technology. She directs the research programs in the Distributed Data Intensive Systems Lab (DiSL), examining various aspects of big data-powered artificial intelligence (AI) systems, and machine learning (ML) algorithms and analytics, including performance, availability, privacy, security, and trust. Prof. Liu is an elected IEEE Fellow, a recipient of IEEE Computer Society Technical Achievement Award (2012), and a recipient of the best paper award from numerous top venues, including IEEE ICDCS, WWW, ACM ⁄ IEEE CCGrid, IEEE Cloud, IEEE ICWS. Prof. Liu served on editorial board of over a dozen international journals, including the editor in chief of IEEE Transactions on Service Computing (2013-2016). Currently, Prof. Liu is the editor in chief of ACM Transactions on Internet Computing (since 2019) and the chair of IEEE CS Fellow Evaluation Committee (FY2024). Prof. Liu is a frequent keynote speaker in top-tier venues in Big Data systems, AI ⁄ ML systems and applications, Cloud Computing, Services Computing, Privacy, Security and Trust. Her current research is primarily supported by USA National Science Foundation under CISE programs, CISCO and IBM. |
