Select a calendar:
Filter May Events by Event Type:
University Calendar
Events for May
-
PhD Thesis Proposal - Rajrup Ghosh
Fri, May 02, 2025 @ 08:30 AM - 10:00 AM
Thomas Lord Department of Computer Science
University Calendar
Title of Thesis Proposal: Enabling Volumetric Video Conferencing and Live Streaming
Date and Time: May 2 2025 (Monday), 8:30 am - 10:00 am PST
Location: 202C, GCS, Floor 2
Committee Members: Ramesh Govindan (Chair), Barath Raghavan, Yue Wang, Harsha V. Madhyastha, Antonio Ortega
Abstract: Volumetric video streaming represents the next frontier in media communication, enabling highly immersive experiences by capturing and transmitting dynamic 3D scenes in real-time. However, significant challenges remain before widespread adoption is possible, including managing substantial bandwidth demands, achieving low-latency, supporting multi-party interactions, and maintaining high visual realism. This thesis explores novel methods to overcome these barriers. To enable immersive two-party conferencing, I introduce LiVo, a system that efficiently streams full-scene volumetric videos by extending 2D video encoding techniques, adaptive bandwidth allocation, and real-time view prediction and culling. LiVo significantly reduces bandwidth usage while maintaining an end-to-end latency of approximately 250 ms at 30 frames per second. Extending these capabilities to multi-party scenarios, I propose LiVo++, which must address challenges in synchronization, computation overhead, and changing network conditions. It proposes dynamic strategies to adapt video quality levels according to participants' device capabilities and available bandwidth, enabling robust and scalable interactive experiences. Finally, I propose to improve visual quality through LiVoGS, a system leveraging Gaussian Splatting. LiVoGS will explore integrating Gaussian Splatting with motion-compensated encoding inspired by traditional 2D codecs, to achieve photorealistic visual quality and significantly improved bandwidth efficiency. Collectively, these innovations enable practical, efficient, and visually realistic volumetric video conferencing and streaming, paving the way toward the future of immersive multimedia communication.
Zoom Link: https://usc.zoom.us/j/92046695163?pwd=PcSYsbDor5695TXOC9en6H6ByMPwbW.1Location: Ginsburg Hall (GCS) - 202C
Audiences: Everyone Is Invited
Contact: Rajrup Ghosh
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Bingjie Tang
Mon, May 05, 2025 @ 11:00 AM - 01:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: From simulation to reality: advancing policy learning for fine-grained manipulation
Date/Time: 05/05/2025 - 11am-1pm.
Venue: 302C in GCS Floor 3
Names of the Dissertation Defense Committee Members:
Gaurav Sukhatme
Jesse Thomason
Erdem Biyik
Feifei Qian
Yashraj Narang
Abstract: Fine-grained manipulation broadly refers to the capability of robots to handle and manipulate objects with high precision, adaptability and robustness. It remains a long-standing challenge in robotics as it includes high-dimensional control, contact-rich dynamics, and robustness towards environment uncertainty. Simulation is a powerful computational tool that has been used for decades to test and optimize safety-critical designs and algorithms in multiple industries. We discuss how we can leverage simulation to enable fast and effective policy learning for fine-grained manipulation and robust transfer to real-world deployment.Location: Ginsburg Hall (GCS) - 302C
Audiences: Everyone Is Invited
Contact: Bingjie Tang
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Sasha Volokh
Tue, May 06, 2025 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Dissertation Title: Using Program Analysis to Determine Actions for Video Game Testing
Date/Time: Tuesday, May 6th, 10:00am-12:00pm
Location: Ginsburg Hall (GCS) - 402C - 4th Floor
Committee: William G.J. Halfond (chair), Nenad Medvidovic, Chao Wang, Mukund Raghothaman, Andrew Nealen
Abstract: In the competitive video game market, the quality of games released to consumers is crucial to their success. However, modern games often release with significant bugs, causing consumer dissatisfaction and a loss of business and reputation for the companies involved. Testing is a key mechanism by which such issues can be caught and addressed during the development process. Many testing approaches require a model of the game rules, which is not available by default for games built with typical game development practices. This poses a barrier to the adoption of more advanced testing techniques, requiring either an expert to model the game or a reliance upon imprecise generic models. At a minimum, knowledge of the possible player actions is crucial for thorough manual and automated testing, but determining a precise and complete model of the game actions is challenging for games built with typical game development practices. In my dissertation, I address these challenges through novel program analysis techniques capable of determining precise and complete models of the game actions. To demonstrate the effectiveness of the proposed techniques, I adapted them to specify the action spaces of automated game testing agents, as well as to generate instructions for assisting human testers. The results show that the action models determined via program analysis enable effective automated testing agent performance and are also capable of improving the exploratory testing performance of human testers.Location: Ginsburg Hall (GCS) - 402C
Audiences: Everyone Is Invited
Contact: Sasha Volokh
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Jesse Zhang
Tue, May 06, 2025 @ 05:00 PM - 06:30 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Scaling Robot Adaptation with Large Model Guidance
Date: Tuesday, May 6, 2025
Time: 5:00 pm - 6:30 pm Location: RTH (Ronald Tutor Hall), room 306
Defense Committee Members: Erdem Biyik (Chair), Jesse Thomason, Joseph J. Lim, Feifei Qian, Gaurav Sukhatme
Abstract: General-purpose robots deployed in the real world must respond to dynamic environments and continuously learn new tasks. However, existing methods struggle to support such adaptation at scale—that is, without substantial human supervision. In this talk, I present an approach to scalable robot adaptation by leveraging the general knowledge encoded in Large Pre-trained Models (LPTMs). I show how integrating LPTMs with robot learning frameworks can: (1) enhance robot pre-training to better prepare for unfamiliar tasks and settings, (2) adapt to new tasks and environments with human feedback, and (3) ultimately enable autonomous adaptation with minimal human input. Together, these contributions outline a path toward generalizable algorithms that empower robots to learn novel tasks in real-world, unstructured environments.Location: Ronald Tutor Hall of Engineering (RTH) - 306
Audiences: Everyone Is Invited
Contact: Ellecia Williams
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Alexander Bisberg
Thu, May 08, 2025 @ 02:15 AM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Modeling Competitive and Social Success in Multiplayer Online Games
Date and Time: Thursday, May 8th, 2025 | 2:15pm
Location: GCS 502C
Committee Members: Emilio Ferrara (Chair), Luca Luceri, Dmitri Williams
Abstract:
This dissertation examines the continuum from competitive to social success in online multiplayer games, employing quantitative and qualitative methods to analyze player behavior and performance across diverse virtual environments. Beginning with a framework for systematically comparing skill rating models in competitive contexts (FRAGEM-S), the research progresses to novel applications of graph neural networks for win prediction, demonstrating improved accuracy and cross-league generalizability. At the intersection of competitive and social domains, analysis of communication patterns in World of Tanks clan networks reveals that high-performing teams exhibit distinctive communication structures characterized by distributed connectivity rather than mere volume. Moving toward the social end of the spectrum, a quasi-experimental study in Sky: Children of Light provides compelling evidence for both generalized reciprocity and third-party influence in virtual worlds, showing that experiencing or witnessing generosity significantly increases future prosocial behavior and game engagement. Finally, unsupervised learning techniques identify persistent behavioral archetypes—including Lone Wolves, Newbies, and Socialites that remain consistent across different time periods and game genres. Together, these findings provide a comprehensive framework for understanding competitive and social success in online games, with implications extending to virtual teams and online communities more broadly.Location: Ginsburg Hall (GCS) - 502C
Audiences: Everyone Is Invited
Contact: Alex Bisberg
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Emily Chen
Thu, May 08, 2025 @ 09:30 AM - 11:30 AM
Thomas Lord Department of Computer Science
University Calendar
Title: Human Behavior in Systems that Undergo Change
Date and Time: Thursday, May 8th, 2025 | 9:30 AM - 11:30 AM
Location: GCS 202C
Committee Members: Emilio Ferrara (Chair), Dmitri Williams, Fred Morstatter
Abstract: In an increasingly digital world, understanding human behavior requires looking at both what people say and what they do online. This dissertation bridges these two dimensions by examining how individuals express themselves on social media and how they behave in virtual game environments.
I first focus on explicit expression, analyzing Twitter data from the COVID-19 pandemic and the 2020 U.S. presidential election to explore how misinformation spreads and how political polarization shapes discourse. These studies show how crises intensify misinformation, particularly within echo chambers.
My research then turns to behavior, using data from two games -- League of Legends and Teamfight Tactics -- to investigate how players respond to different structural incentives. Despite the games rewarding different strategies, individual behavior remains surprisingly stable, highlighting the persistent influence of agency.
Together, these studies offer a multi-modal perspective on online behavior, contributing to computational social science and informing the design and governance of sociotechnical platforms.Location: Ginsburg Hall (GCS) - 202C
Audiences: Everyone Is Invited
Contact: Emily Chen
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Elan Markowitz
Thu, May 08, 2025 @ 02:30 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Integrating Knowledge Graphs and Large Language Models to Improve Factuality and Reasoning
Date and Time: Thursday, May 8th, 2025 | 2:30p Location: GCS 202C
Committee Members: Aram Galstyan, Greg Ver Steeg, Bistra Dilkina, Antonio Ortega
Abstract: Large Language Models (LLMs) have rapidly emerged as the dominant paradigm in AI due to their powerful understanding of unstructured text, strong reasoning abilities, and highly general task completion capabilities. However, they also have limitations in terms of how they use knowledge. They are black boxes with internal reasoning that is hard to analyze; they hallucinate incorrect facts as if they are true; and they suffer from knowledge cutoffs based on when their training ends. Knowledge graphs naturally complement these weaknesses through providing vast, structured, up-to-date, information over both general and specific domains. At the same time, knowledge graphs have limitations, such as incompleteness and limited reasoning, that can be complemented by Large Language Models. Ultimately, through better integrating these approaches, we will deliver more reliable and trustworthy AI systems.
In this dissertation, I present a body of research on combining Large Language Models and Knowledge Graphs to address many of their individual weaknesses. This includes topics such as addressing knowledge graph incompleteness through combining language models and more structured graph neural networks; Integrating LLMs and external knowledge graphs with advanced reasoning capabilities; Measuring how presentation and other factors impact Large Language Models' understanding of in-context Knowledge Graphs; and using Knowledge Graphs to improve model editing approaches for updating an LLM’s internal knowledge.Location: Ginsburg Hall (GCS) - 202C
Audiences: Everyone Is Invited
Contact: Elan Markowitz
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Zhaoxu Zhang
Fri, May 09, 2025 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Dissertation Title: Automated Reproduction of Bug Reports for Mobile Applications
Date/Time: Friday, May 9th, 10:00 am-12:00 pm
Location: Ginsburg Hall (GCS) - 202C - 2nd Floor Location: GCS 202C
Committee: William G.J. Halfond (chair), Nenad Medvidovic, Chao Wang, Jesse Thomason, and Sandeep Gupta.
Abstract:Mobile app developers need to reproduce the failures described in bug reports submitted by app users in order to fix the bug. However, due to the often low quality of bug reports and the complexity of modern applications, this manual reproduction process can be challenging and time-consuming. As a result, there is a significant demand for automated solutions that can assist in reproducing mobile app bug reports. Unfortunately, existing methods for reproducing mobile app bug reports have several limitations. They typically handle only limited forms of natural language text in the bug report, struggle to reproduce bugs when the report lacks accurate and complete reproduction steps, and are unable to reproduce non-crash bugs. In my dissertation, I developed and implemented several techniques to address the limitations of existing approaches and enhance the automated reproduction process for mobile app bug reports. First, I developed an approach that leverages a set of Natural Language Processing analyses to extract step information from bug reports, handling a wider variety of text than existing methods. Second, I introduced two algorithms designed to identify UI events to reproduce the reproduction steps, specifically aimed at addressing the challenges posed by incomplete and inaccurate steps. Third, I designed an approach that automatically recognizes buggy behaviors based on bug reports, enabling the automated reproduction of non-crash bug reports. I evaluated the effectiveness of each technique using real-world bug reports and assessed the overall reproduction performance by integrating them into end-to-end reproduction approaches. The results demonstrated that each individual technique achieved high accuracy, and the combined reproduction approach significantly outperformed state-of-the-art approaches in reproducing mobile app bug reports.Location: Ginsburg Hall (GCS) - 202C
Audiences: Everyone Is Invited
Contact: Zhaoxu Zhang
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Thesis Proposal - James Huang
Fri, May 09, 2025 @ 03:00 PM - 04:30 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Collaborative Decision-Making of Language Models
Date and Time: Friday, May 9th, 2025 | 3:00p - 4:30p
Location: GCS 502C
Committee Members: Muhao Chen, Fred Morstatter, Laurent Itti, Robbin Jia, Dan O'Leary
Abstract: While general-purpose language models have demonstrated strong performance on a wide range of tasks, they still have their own weaknesses such as biases, misalignment, lack of task-specific knowledge, etc. One promising way of addressing these challenges is to combine the strengths of different language models. In this proposal, I will outline my research exploring various strategies to facilitate collaborative decision making of language models. Specifically, I will present 1) a shortcut mitigation method via ensemble-based attention debiasing, 2) a decoding-time alignment framework that uses model-based reward functions to guide model generation, and 3) an unlearning method that removes sensitive knowledge by learning a logit offset. Finally, I will discuss future directions for language model collaboration.Location: Ginsburg Hall (GCS) - 502C
Audiences: Everyone Is Invited
Contact: James Huang
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
TIEHub YCombinator Application Workshop & Mixer
Fri, May 09, 2025 @ 05:00 PM - 06:00 PM
Viterbi Technology Innovation and Entrepreneurship
University Calendar
Meet other USC YC applicants & get feedback on your application over pizza!
Location: Michelson Center for Convergent Bioscience (MCB) - 102
Audiences: Everyone Is Invited
Contact: Viterbi TIE
Event Link: https://cglink.me/2nB/r404355
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Thesis Proposal - Christina Shin
Tue, May 13, 2025 @ 11:00 AM - 12:30 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Harnessing Vehicular 3D Primitives for Networked Scene Operation and Manipulation
Date and Time: Tuesday, May 13th, 2025 | 11am-12:30pm
Location: GCS 402C
Committee Members: Ramesh Govindan (Advisor), Laurent Itti, Harsha V. Madhyastha, Antonio Ortega, Barath Raghavan
Abstract: This proposal explores a cloud-centric framework for managing 3D data as a networked and shared resource in vehicular systems. Fueled by 3D sensors like LiDARs, 3D data enables new ways of perceiving and interacting with the physical world with high spatial fidelity. With the rise of vehicular communication, 3D data can now be streamed, fused, and interpreted beyond local on-board devices enabling new forms of collaborative scene understanding and immersive content delivery.
The proposal introduces two systems: RECAP, which reconstructs traffic scenes by aggregating 3D data from moving vehicles, and CIP, which performs collaborative perception using multiple infrastructure sensors. These works demonstrate how cloud processing can support real-time, accurate, and scalable 3D scene operations. Looking ahead, the proposed system MARS aims to deliver 3D video to vehicles for immersive passenger experiences, expanding the use of 3D data beyond machine perception to human-centered applications.
Location: Ginsburg Hall (GCS) - 402C
Audiences: Everyone Is Invited
Contact: Christina Shin
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Paul Chiou
Tue, May 13, 2025 @ 12:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Automated Detection of Keyboard Accessibility Issues in Web Applications
Date and Time: Tuesday, May 13th, 2025. 12:00 PM - 2:00 PM
Location: Ginsburg Hall (GCS) Floor L5 - Room 502C
Committee Members: William G.J. Halfond (Chair), Nenad Medvidovic, Mukund Raghothaman, Gisele Ragusa, and Chao Wang
Abstract: The internet has become an important part of our daily lives, enabling us to complete everyday and essential tasks online. For the 15% of the global population with disabilities, accessing the internet is critical and can provide access to resources that would otherwise be unavailable. Many people with different disabilities rely on the keyboard interface to access the internet; however, studies found that web applications today largely remain inaccessible to keyboard users. Testing keyboard accessibility is a labor-intensive task currently done manually by skilled practitioners. In my research, I used program analysis techniques to automate the keyboard accessibility testing process to alleviate the manual effort involved. I developed a novel approach to automatically detect keyboard accessibility issues that negatively affect disabled users' ability to navigate web pages' user interface. The approach implements a dynamic crawler to build a model that captures a web page's interactivity from a keyboard user's perspective. The approach then analyzes the model to identify the inaccessible behaviors per accessibility guidelines. Finally, I conducted evaluations to show the accuracy of the approach in detecting keyboard accessibility issues in real-world web applications.Location: Ginsburg Hall (GCS) - 502C
Audiences: Everyone Is Invited
Contact: Paul Chiou
Event Link: https://usc.zoom.us/j/94969307418?pwd=tEEvSPznMZgr7DBEvP4T5vREfBCYD0.1
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Hsien-Te Kao
Tue, May 13, 2025 @ 02:00 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Cold Start Prediction in Personalizd Health
Date and Time: Tuesday, May 13th, 2025 | 2:00p - 4:00p
Location: GCS 202C
Committee Members: Emilio Ferrara (Chair), Kristina Lerman, Maryam M. Shanechi
Abstract: Mobile health (mHealth) has transformed healthcare delivery by using mobile technologies, wearable sensors, and machine learning to expand access, especially for populations facing geographic, economic, or clinical barriers. By enabling passive and continuous data collection, mHealth systems support early detection, real-time prediction, and proactive management of a wide range of health conditions through sensor-driven machine learning. Personalized mHealth extends these capabilities by integrating individual-level modeling and multi-source health records to improve model performance and support deeper understanding of individual health in their life contexts. Despite this progress, real-world deployment remains constrained by user reluctance, privacy concerns, and strict regulations that severely limit the availability of labeled individual health data. This dissertation presents a personalized mHealth framework designed to achieve mHealth predictions without health labels, addressing the cold-start problem. The work identifies key temporal segments that most influence model performance, introduces a cognitive appraisal-based similarity metric linking individuals through physiological signals and health labels, and demonstrates that five labels are sufficient for assigning users into their appraisal cohorts. It further shows that promising mHealth predictions can be achieved under cold-start conditions and uncovers how sociodemographic factors are associated with latent physiological and health patterns. The research contributes to foundational advances in theory-driven, label-efficient modeling for individualized health prediction. It also supports the development of practical mHealth systems capable of improving everyday health management beyond clinical settings.Location: Ginsburg Hall (GCS) - 202C
Audiences: Everyone Is Invited
Contact: Hsien-Te Kao
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
PhD Dissertation Defense - Soumya Sanyal
Fri, May 16, 2025 @ 02:00 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Demystifying and Improving Large Language Models on Consistent Reasoning
Date and Time: Friday, May 16th, 2025 | 2:00p - 4:00p
Location: GCS 402C
Committee Members: Xiang Ren (Chair), Robin Jia, Morteza Deghani
Abstract: Large Language Models (LLMs) have demonstrated remarkable performance on various language tasks. However, their reasoning processes often lack consistency, a fundamental requirement for trust, reliability, and interpretability. In this talk, I define reasoning consistency as a multi-faceted concept that requires logical coherence, non-contradiction, and robustness to semantic and structural perturbations. My research focuses on three core areas: (1) benchmarking the consistency of LLMs on deductive reasoning tasks, (2) detecting inconsistencies in LLMs' internal reasoning across different language tasks, and (3) developing techniques to enhance the consistency and reliability of LLM reasoning. The findings aim to understand the behavior of LLMs and enhance their reliability in real-world applications where consistent reasoning is critical.Location: Ginsburg Hall (GCS) - 402C
Audiences: Everyone Is Invited
Contact: Somuya Sanyal
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor.
