Computer Science Theses and Dissertations

Permanent URI for this collection

https://hdl.handle.net/1794/2040

This collection contains some of the theses and dissertations produced by students in the University of Oregon Computer Science Graduate Program. Paper copies of these and other dissertations and theses are available through the UO Libraries.

Browse

Browsing Computer Science Theses and Dissertations by Subject "Adversarial Learning"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Improving Cross-Lingual Transfer Learning for Event Detection
    (University of Oregon, 2024-01-09) Guzman Nateras, Luis; Nguyen, Thien
    The widespread adoption of applications powered by Artificial Intelligence (AI) backbones has unquestionably changed the way we interact with the world around us. Applications such as automated personal assistants, automatic question answering, and machine-based translation systems have become mainstays of modern culture thanks to the recent considerable advances in Natural Language Processing (NLP) research. Nonetheless, with over 7000 spoken languages in the world, there still remain a considerable number of marginalized communities that are unable to benefit from these technological advancements largely due to the language they speak. Cross-Lingual Learning (CLL) looks to address this issue by transferring the knowledge acquired from a popular, high-resource source language (e.g., English, Chinese, or Spanish) to a less favored, lower-resourced target language (e.g., Urdu or Swahili). This dissertation leverages the Event Detection (ED) sub-task of Information Extraction (IE) as a testbed and presents three novel approaches that improve cross-lingual transfer learning from distinct perspectives: (1) direct knowledge transfer, (2) hybrid knowledge transfer, and (3) few-shot learning.
  • Thumbnail Image
    ItemOpen Access
    Limitations of Data-Based Decision Making: An Multiparadigmic Investigation of Challenges Faced by Artificial Intelligence
    (University of Oregon, 2025-02-24) Kinsey, Sarah; Nguyen, Thanh
    As big data and computing capability continue to grow, an ever-increasingamount of artificial intelligence approaches are being deployed in the real world, across various domains. With real world deployments come additional complexities, challenges, and vulnerabilities, particularly concerning data reliance. Among others, these data related vulnerabilities include the potential for intelligently sabotaged data, and incomplete data. This work explores vulnerabilities from three perspectives: adversarial learning, game theory, and online reinforcement learning. First, we investigate whether a directly targeted end-to-end poisoning attack on a data-based decision making learning-planning model is feasible. Next, through the lens of security games, we investigate how a data-based decision maker can form a useful behavioral model, despite the observable data being maliciously manipulated. Lastly, we examine how, in a real world online RL setting, limited and sparse data can be overcome to build an effective data-based decision making model.