Logo: University of Southern California

Events Calendar



Select a calendar:



Filter March Events by Event Type:



University Calendar
Events for March

  • PHD Thesis Defense - Dimitris Stripelis

    Mon, Mar 20, 2023 @ 11:00 AM - 12:00 PM

    Thomas Lord Department of Computer Science

    University Calendar


    PHD Thesis Defense - Dimitris Stripelis

    Title:
    Heterogeneous Federated Learning

    Committee Members:
    Jose-Luis Ambite (Chair), Cyrus Shahabi, Paul Thompson, Greg Ver Steeg


    Abstract:
    Data relevant to machine learning problems are distributed across multiple data silos that cannot share their data due to regulatory, competitiveness, or privacy reasons. Federated Learning has emerged as a standard computational paradigm for distributed training of machine learning and deep learning models across silos. However, the participating silos may have heterogeneous system capabilities and data specifications. In this thesis, we address the challenges in federated learning arising from both computational and semantic heterogeneities. We present federated training policies that accelerate the convergence of the federated model and lead to reduced communication, processing, and energy costs during model aggregation, training, and inference. We show the efficacy of these policies across a wide range of challenging federated environments with highly diverse data distributions in benchmark domains and in neuroimaging. We conclude by describing the federated data harmonization problem and presenting a comprehensive federated learning and integration system architecture that addresses the critical challenges of secure and private federated data harmonization, including schema mapping, data normalization, and data imputation.

    Location: https://usc.zoom.us/j/93599773555?pwd=TmI3M1JvTkxEV05DSmQ3dzYyVElmQT09

    Audiences: Everyone Is Invited

    Contact: Asiroh Cham

    Event Link: https://usc.zoom.us/j/93599773555?pwd=TmI3M1JvTkxEV05DSmQ3dzYyVElmQT09

    OutlookiCal
  • PhD Thesis Proposal - Gautam Salhotra

    Thu, Mar 23, 2023 @ 09:00 AM - 11:00 AM

    Thomas Lord Department of Computer Science

    University Calendar


    Title: Accelerating Robot Reinforcement Learning Using Demonstrations

    Committee: Gaurav Sukhatme (Chair), SK Gupta, Laurent Itti, Stefanos Nikolaidis, Somil Bansal

    Date: Thursday March 23, 9am PST

    Abstract:
    Reinforcement learning is a promising and, recently, popular tool to solve robotic tasks such as object manipulation and locomotion. However, it is also well known for being a very hard problem setting to explore in. In contrast, Learning from demonstrations (LfD) methods train agents to the desired solution using demonstrations from a teacher.
    I will explore the role of LfD methods to guide the exploration of RL methods, with the aim of applying it to regular object manipulation tasks. I will talk about work that uses planners and trajectory optimizers to guide RL, and then discuss the role human experts can play in LfD for RL. Finally, I will talk about proposed projects that can extend the current work to get the benefits of demonstrations while avoiding the downsides of obtaining them.



    Location: Ronald Tutor Hall of Engineering (RTH) - 406

    Audiences: Everyone Is Invited

    Contact: Melissa Ochoa

    OutlookiCal
  • PhD Thesis Proposal - Basel Shbita

    Mon, Mar 27, 2023 @ 10:30 AM - 12:00 PM

    Thomas Lord Department of Computer Science

    University Calendar


    Title:
    Automatic Semantic Spatio-Temporal Interpretation of Historical Maps

    Committee:
    Craig A. Knoblock (chair), Cyrus Shahabi, John P. Wilson, Jay Pujara, Yao-Yi Chiang

    Date:
    Friday, February 17th, 1:30pm-3pm PST

    Zoom Meeting Details:
    https://usc.zoom.us/j/97387539087?pwd=MWEwaHR0Z0FCOEdwdGdEcWxFSnorZz09
    Meeting ID: 973 8753 9087
    Passcode: 312501

    Abstract:
    Historical maps provide rich information for researchers in many areas, including the natural and social sciences. These maps include detailed documentation of a wide variety of natural and human-made features, their spatial extent, their changes over time, their geo-names, and additional metadata. Analyzing map collections that cover the same region at different points in time can be labor-intensive even for a scientist, often requiring further grounding and linking with external sources to contextualize the data. With rapidly increasing amounts of digitized map archives, we require methods to convert these maps into a machine-processable and machine-readable semantic form and do so automatically, efficiently, and at scale. Unfortunately, existing techniques are limited and do not leverage the vast landscape of information extracted from archives of historical maps.
    In this thesis proposal, we investigate how to convert the extracted geo-data and metadata to a dynamic knowledge graph representation that captures the data semantics, how the data can be interrelated across entire datasets, and how it can be grounded to real-world phenomena by leveraging external resources on the web. We explore approaches that benefit from the open and connective nature of linked data that can produce a spatio-temporal, semantic, and contextualized output that follows linked data principles, and that can be easily extended with further availability of contemporary maps while supporting backward compatible access. Once materialized in a dynamic knowledge graph, the output can hold the data in a semantic network, making it readily shared, accessible, visualized, standardized across domains, and scalable for effortless use by downstream tasks for analysis and expressive integration over time and space.

    WebCast Link: https://usc.zoom.us/j/97387539087?pwd=MWEwaHR0Z0FCOEdwdGdEcWxFSnorZz09

    Audiences: Everyone Is Invited

    Contact: Asiroh Cham

    OutlookiCal
  • 2023 Mork Family Department Student Research Symposium

    Fri, Mar 31, 2023 @ 09:00 AM - 03:30 PM

    Mork Family Department of Chemical Engineering and Materials Science

    University Calendar


    Location: Michelson Center for Convergent Bioscience (MCB) - First Floor

    Audiences: Everyone Is Invited

    Contact: Candy Escobedo

    OutlookiCal