Select a calendar:
Filter May Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for May
-
PhD Defense- Woojeong Jin
Wed, May 01, 2024 @ 10:00 AM - 11:30 AM
Thomas Lord Department of Computer Science
Student Activity
PhD Defense- Woojeong Jin
Title: Bridging the Visual Knowledge Gaps in Pre-trained Models
Committee: Xiang Ren (chair), Ram Nevatia, Yan Liu, Toby Mintz.
Abstract: Humans acquire knowledge by processing visual information through observation and imagination, which expands our reasoning capability about the physical world we encounter every day. Despite significant progress in solving AI problems, current state-of-the-art models in natural language processing (NLP) and computer vision (CV) have limitations in terms of reasoning and generalization, particularly with complex reasoning on visual information and generalizing to unseen vision-language tasks. In this thesis, we aim to build a reasoner that can do complex reasoning about the physical world and generalization on vision-language tasks. we will present a few lines of work to bridge the visual knowledge gaps in pre-trained models.Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Woojeong Jin
-
PhD Dissertation Defense- Basel Shbita
Wed, May 01, 2024 @ 03:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Student Activity
PhD Dissertation Defense- Basel Shbita
Title: Transforming Unstructured Historical and Geographic Data into Spatio-Temporal Knowledge Graphs
Committee: Craig A. Knoblock (chair), Cyrus Shahabi, John P. Wilson; Jay Pujara, Yao-Yi Chiang
Abstract: This dissertation presents a comprehensive approach to the transformation, integration and semantic enrichment of historical spatio-temporal data into knowledge graphs. The dissertation encompasses three core contributions: one, the automated generation of knowledge graphs from digitized historical maps for analyzing geographical changes over time; two, the integration of spatial and semantic context embeddings for accurate geo-entity recognition and semantic typing; and three, the creation of a comprehensive knowledge graph for the analysis of historical data from digitized archived records. I introduce innovative methodologies and practical tools to support researchers from diverse fields, enabling them to derive meaningful insights from historical and geographic data. My approach is demonstrated through various applications, such as analyzing geospatial changes over time in USGS (United States Geological Survey) historical maps of transportation networks and wetlands, automatic semantic typing of unlabeled georeferenced spatial entities, and constructing a spatio-temporal knowledge graph from digitized historical mineral mining data. The dissertation combines semantic web technologies, representation learning, and semantic modeling to build comprehensive knowledge graphs that support geospatial and temporal analyses.Audiences: Everyone Is Invited
Contact: Basel Shbita
Event Link: https://usc.zoom.us/j/97894910088?pwd=ZVQ0VU9lYlJaWTM4V2w5Vk1maEVOQT09
-
PhD Thesis Proposal - Ta-Yang Wang
Wed, May 01, 2024 @ 03:00 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Training Heterogeneous Graph Neural Networks using Bandit Sampling
Presenter: Ta-Yang Wang
Time: May 1st, 3:00 PM - 4:00 PM
Location: EEB 219
Committee members: Viktor Prasanna (chair), Jyotirmoy Deshmukh, Rajgopal Kannan, Aiichiro Nakano, and Cauligi Raghavendra
Abstract: Graph neural networks (GNNs) have gained significant attention across diverse areas due to their superior performance in learning graph representations. While GNNs exhibit superior performance compared to other methods, they are primarily designed for homogeneous graphs, where all nodes and edges are of the same type. Training a GNN model for large-scale graphs incurs high computation and storage costs, especially when considering the heterogeneous structural information of each node. To address the demand for efficient GNN training, various sampling methods have been proposed. In this proposal, we hypothesize that one can improve the training efficiency via bandit sampling, an online learning algorithm with provable convergence under weak assumptions on the learning objective. The main idea is to prioritize node types with more informative connections with respect to the learning objective. Additionally, we analyze the limitations of the framework, thus advancing its applicability in large-scale graph learning tasks.Location: Hughes Aircraft Electrical Engineering Center (EEB) - 219
Audiences: Everyone Is Invited
Contact: Ellecia Williams
-
Machine Learning Center Seminar
Thu, May 02, 2024 @ 12:00 PM - 01:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Pengtao Xie , Assistant Professor, Department of Electrical and Computer Engineering - University of California, San Diego
Talk Title: Foundation Models and Generative AI for Medical Imaging Segmentation in Ultra-Low Data Regimes
Abstract: Semantic segmentation of medical images is pivotal in disease diagnosis and treatment planning. While deep learning has excelled in automating this task, a major hurdle is the need for numerous annotated masks, which are resource-intensive to produce due to the required expertise and time. This scenario often leads to ultra-low data regimes where annotated images are scarce, challenging the generalization of deep learning models on test images. To address this, we introduce two complementary approaches. One involves developing foundation models. The other involves generating high-fidelity training data consisting of paired segmentation masks and medical images. In the former, our bi-level optimization based method can effectively adapt the general-domain Segment Anything Model (SAM) to the medical domain with just a few medical images. In the latter, our multi-level optimization based method can perform end-to-end generation of high-quality training data from a minimal number of real images. On eight segmentation tasks involving various diseases, organs, and imaging modalities, our methods demonstrate strong generalization performance in both in-domain and out-of-domain settings. Our methods require 8-12 times less training data than baselines to achieve comparable performance.
Biography: Pengtao Xie is an assistant professor in the Department of Electrical and Computer Engineering at the University of California San Diego. His research interest lies in machine learning for healthcare. His PhD thesis was selected as a top-5 finalist for the Doctoral Dissertation Award of the American Medical Informatics Association (AMIA). He was recognized as Global Top-100 Chinese Young Scholars in Artificial Intelligence by Baidu, Tencent AI-Lab Faculty Award, Innovator Award by the Pittsburgh Business Times, Amazon AWS Machine Learning Research Award, among others. He serves as an associate editor for the ACM Transactions on Computing for Healthcare, senior area chair for AAAI, area chairs for ICML and NeurIPS, etc.
Host: Machine Learning Center
Location: Ronald Tutor Hall of Engineering (RTH) - 211
Audiences: Everyone Is Invited
Contact: Thomas Lord Department of Computer Science
-
PhD Thesis Defense - Matthew Ferland
Thu, May 02, 2024 @ 12:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Thesis Defense: Matthew Ferland
Committee: Shanghua Teng (Chair), David Kempe, Jiapeng Zhang, Larry Goldstein (Math)
Title: Exploring the Computational Frontier of Combinatorial Games
Abstract: People have been playing games since before written history, and many of the earliest games were combinatorial games, that is to say, games of perfect information and no chance. This type of game is still widely played today, and many popular games of this type, such as Chess and Go, are some of the most studied games of all time. This proposed work resolves around a game-independent systemic study of these games. More specifically, computational properties involving evaluating mathematical analysis tools for combinatorial games, such as Grundy values and confusion intervals, as well as identifying what can be determined about these games using simple oracle models.Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: CS Events
-
PhD Defense- Julie Jiang
Fri, May 03, 2024 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Student Activity
PhD Defense- Julie Jiang
Title: Socially-informed content analysis of online human behavior
Committee: Emilio Ferrara (CS and Communication, tenure, chair), Kristina Lerman (CS), Marlon Twyman II (Communication, external), Pablo Barberá (Poli Sci)
Abstract: The explosive growth of social media has not only revolutionized communication but also brought challenges such as political polarization, misinformation, hate speech, and echo chambers. This dissertation employs computational social science techniques to investigate these issues, understand the social dynamics driving negative online behaviors, and propose data-driven solutions for healthier digital interactions. I begin by introducing a scalable social network representation learning method that integrates user-generated content with social connections to create unified user embeddings, enabling accurate prediction and visualization of user attributes, communities, and behavioral propensities. Using this tool, I explore three interrelated problems: 1) COVID-19 discourse on Twitter, revealing polarization and asymmetric political echo chambers; 2) online hate speech, suggesting the pursuit of social approval motivates toxic behavior; and 3) moral underpinnings of COVID-19 discussions, uncovering patterns of moral homophily and echo chambers, while also indicating moral diversity and plurality can improve message reach and acceptance across ideological divides. These findings contribute to the advancement of computational social science and provide a foundation for understanding human behavior through the lens of social interactions and network homophily.
Location: Grace Ford Salvatori Hall Of Letters, Arts & Sciences (GFS) - 104
Audiences: Everyone Is Invited
Contact: Julie Jiang
Event Link: https://usc.zoom.us/j/5152754393?pwd=V1pzUnpEc0JtTVZlS0l5R1VMRWlRdz09&omn=91709345144
-
PhD Defense- Yilei Zeng
Tue, May 07, 2024 @ 10:30 AM - 12:00 PM
Thomas Lord Department of Computer Science
Student Activity
PhD Defense- Yilei Zeng
Title: Learning Social Sequential Decision-Making in Online Games
Committee: Emilio Ferrara (chair), Dmitri Williams, Michael Zyda
Abstract:
A paradigm shift towards human-centered intelligent gaming systems is gradually setting in. This dissertation explores the complexities of social sequential decision-making within online gaming environments and presents comprehensive AI solutions to enhance personalized single and multi-agent experiences. The three core contributions of the dissertation are intricately interrelated, creating a cohesive framework for understanding and improving AI in gaming. I begin by delving into the dynamics of gaming sessions and sequential in-game individual and social decision-making, which establishes a baseline of how decisions evolve, providing the necessary context for the subsequent integration of diverse information sources; two, the integration of heterogeneous information and multi-modal trajectories, which enhances decision-making generation models; and three, the creation of a reinforcement learning with human feedback framework to train gaming AIs that effectively align with human preferences and strategies, which enables the system not only learning but also interacting with humans. Collectively, this dissertation combines innovative data-driven, generative AI, representation learning, and human-AI collaboration solutions to help advance both the fields of computational social science and artificial intelligence applications of gaming.
Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: Yilei Zeng
-
PhD Dissertation Defense - Yilei Zeng
Tue, May 07, 2024 @ 10:30 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Learning Social Sequential Decision-Making in Online Games
Date and Time: May 7th, 2024 - 10:30a - 12:00p
Location: SAL 213
Committee: Emilio Ferrara (Chair), Dmitri Williams, Michael Zyda
Abstract: A paradigm shift towards human-centered intelligent gaming systems is gradually setting in. This dissertation explores the complexities of social sequential decision-making within online gaming environments and presents comprehensive AI solutions to enhance personalized single and multi-agent experiences. The three core contributions of the dissertation are intricately interrelated, creating a cohesive framework for understanding and improving AI in gaming. I begin by delving into the dynamics of gaming sessions and sequential in-game individual and social decision-making, which establishes a baseline of how decisions evolve, providing the necessary context for the subsequent integration of diverse information sources; two, the integration of heterogeneous information and multi-modal trajectories, which enhances decision-making generation models; and three, the creation of a reinforcement learning with human feedback framework to train gaming AIs that effectively align with human preferences and strategies, which enables the system not only learning but also interacting with humans. Collectively, this dissertation combines innovative data-driven, generative AI, representation learning, and human-AI collaboration solutions to help advance both the fields of computational social science and artificial intelligence applications of gaming.Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: Yilei Zeng
-
PhD Dissertation Defense - I-Hung Hsu
Tue, May 07, 2024 @ 02:10 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Towards Generalized Event Understanding in Text via Generative Models
Committee Members: Dr. Prem Natarajan (Chair), Dr. Nanyun Peng (Co-Chair), Dr. Dan O'Leary, Dr. Emilio Ferrara
Date and Time: May 7th, 2024 - 2:10p - 4:00p
Abstract: Human languages in the world, such as news or narratives, are structured around events. Focusing on these events allows Natural Language Processing (NLP) systems to better understand plots, infer motivations, consequences, and the dynamics of situations. Despite the rapidly evolving landscape of NLP technology, comprehending complex events, particularly those rarely encountered in training such as in niche domains or low-resource languages, remains a formidable challenge. This thesis explores methods to enhance NLP model generalizability for better adaptability to unfamiliar events and languages unseen during training.
My approach includes two main strategies: (1) Model Perspective: I propose a novel generation-based event extraction framework, largely different from typical solutions that make predictions by learning to classify input tokens. This new framework utilizes indirect supervision from natural language generation, leveraging large-scale unsupervised data without requiring additional training modules dependent on limited event-specific data. Hence, it facilitates the models’ ability on understanding general event concepts. I further explore advanced methods to extend this framework for cross-lingual adaptation and to utilize cross-domain robust resources effectively. (2) Data Perspective: I develop techniques to generate pseudo-training data broaden the training scope for event understanding models. This includes translating structured event labels into other languages with higher accuracy and fidelity, and synthesizing novel events for the existing knowledge base.
Overall, my work introduces a novel learning platform to the NLP community, emphasizing an innovative modeling paradigm and comprehensive data preparation to foster more generalized event understanding models.
Location: Information Science Institute (ISI) - 727
Audiences: Everyone Is Invited
Contact: I-Hung Hsu
Event Link: https:/usc.zoom.us/j/95785927723?pwd=dFlGbEcwbXlGalJ6OVk3YW41RDMrdz09
-
PhD Thesis Defense - Qinyi Luo
Thu, May 09, 2024 @ 11:00 AM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Thesis Defense - Qinyi (Chelsea) Luo
Committee members: Xuehai Qian (co-chair), Viktor Prasanna (co-chair), Ramesh Govindan, Chao Wang, Feng Qian
Title: High-Performance Heterogeneity-Aware Distributed Machine Learning Model Training
Abstract: The increasing size of machine learning models and the ever-growing amount of data result in days or even weeks of time required to train a machine learning model. To accelerate training, distributed training with parallel stochastic gradient descent is widely adopted as the go-to training method. This thesis targets four challenges in distributed training: (1) performance degradation caused by large amount of data transfer among parallel workers, (2) heterogeneous computation and communication capacities in the training devices, i.e., the straggler issue, (3) huge memory consumption during training caused by gigantic model sizes, and (4) automatic selection of parallelization strategies. This thesis first delves into the topic of decentralized training and proposes system support and algorithmic innovation that strengthen tolerance against stragglers in data-parallel training. On the system side, a unique characteristic of decentralized training, the iteration gap, is identified, and a queue-based synchronization mechanism is proposed to efficiently support decentralized training as well as common straggler-mitigation techniques. In the experiments, the proposed training protocol, Hop, can provide strong tolerance against stragglers and train much faster than standard decentralized training when stragglers are present. On the algorithm side, a novel communication primitive, randomized partial All-Reduce, is proposed to enable fast synchronization in decentralized data-parallel training. The proposed approach, Prague, can achieve a 1.2x speedup against All-Reduce in a straggler-free environment and a 4.4x speedup when stragglers are present. Then, on the topic of memory optimization for training Deep Neural Networks (DNNs), an adaptive during-training model compression technique, FIITED, is proposed to reduce the memory consumption of training huge recommender models. FIITED adapts to dynamic changes in data and adjusts the dimension of each individual embedding vector continuously during training. Experiments show that FIITED is able to reduce the memory consumption of training significantly more than other embedding pruning methods, while maintaining the trained model's quality. In the end, in the aspect of automatic parallelization of training workloads, a novel unified representation of parallelization strategies, incorporating Data Parallelism (DP), Model Parallelism (MP) and Pipeline Parallelism (PP), is proposed, as well as a search algorithm that selects superior parallel settings in the vast search space. An ideal stage partition ratio for synchronous pipelines is derived for the first time, to the best of my knowledge, and it is theoretically proven that unbalanced partitions are better than balanced partitions. In addition, by examining the pipeline schedule, a trade-off between memory and performance is uncovered and explored. Experiments show that hybrid parallel strategies generated with the aforementioned optimizations consistently outperform those without such considerations.
Date: May 9, 2024
Time: 11:00 a.m. - 1:00 p.m.
Location: EEB 110
Zoom link: https://usc.zoom.us/j/95741130954?pwd=dkRkblNlNGt0TlkwOU51SlRNS0hPZz09Location: Hughes Aircraft Electrical Engineering Center (EEB) -
Audiences: Everyone Is Invited
Contact: CS Events
Event Link: https://usc.zoom.us/j/95741130954?pwd=dkRkblNlNGt0TlkwOU51SlRNS0hPZz09
-
PhD Dissertation Defense - Nicolaas Weideman
Thu, May 16, 2024 @ 09:00 AM - 11:00 AM
Thomas Lord Department of Computer Science
Student Activity
PhD Dissertation Defense - Nicolaas Weideman
Title: Improving Binary Program Analysis to Enhance the Security of Modern Software Systems
Committee: Jelena Mirkovic (chair), Chao Wang and Paul Bogdan
Abstract: With the ever-increasing reliance of the modern world on software systems, the frequency and impact of cyberattacks have greatly increased as well. Software must be analyzed thoroughly to evaluate its security, as vulnerabilities in software can have devastating consequences such as compromised privacy of users, shutdown of infrastructure, significant business losses, and even pose threat to human life. Unfortunately, manual analysis of the source code is insufficient to evaluate the security of software. This is firstly due to the quantity and size of modern software making this method impractical and secondly due to low-level vulnerabilities that are invisible in the source code. Conversely, binary program analysis focuses on automatically analyzing the machine code instructions of executables to reason about security-related properties. In this thesis we enhance automatic software security evaluation by leveraging and extending binary program analysis. We develop approaches to 1) automatically discover vulnerabilities and 2) automatically and safely patch vulnerabilities. We improve the reliability of binary data-flow analysis by 3) evaluating three state of the art binary analysis frameworks and 4) improving the state of the art. Each of these directions independently pushes the boundaries of what is possible in defending modern software, leading to a more secure digital environment.
Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: Nicolaas Weideman
