VTechWorks staff will be away for the Thanksgiving holiday beginning at noon on Wednesday, November 27, through Friday, November 29. We will resume normal operations on Monday, December 2. Thank you for your patience.

Browsing by Author "Li, Xuan"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Referencing Unlabelled World Data to Prevent Catastrophic Forgetting in Class-incremental Learning
    Li, Xuan (Virginia Tech, 2022-06-24)
    This thesis presents a novel strategy to address the challenge of "catastrophic forgetting" in deep continual-learning systems. The term refers to severe performance degradation for older tasks, as a system learns new tasks that are presented sequentially. Most previous techniques have emphasized preservation of existing knowledge while learning new tasks, in some cases advocating a memory buffer that grows in proportion to the number of tasks. However, we offer another perspective, which is that mitigating local-task fitness during learning is as important as attempting to preserve existing knowledge. We posit the existence of a consistent, unlabelled world environment that the system uses as an easily-accessible reference to avoid favoring spurious properties over more generalizable ones. Based on this assumption, we have developed a novel method called Learning with Reference (LwR), which delivers substantial performance gains relative to its state-of-the-art counterparts. The approach does not involve a growing memory buffer, and therefore promotes better performance at scale. We present extensive empirical evaluation on real-world datasets.
  • "Slow down. Rail crossing ahead. Look left and right at the crossing": In-vehicle auditory alerts improve driver behavior at rail crossings
    Nadri, Chihab; Kekal, Siddhant; Li, Yinjia; Li, Xuan; Lee, Seul Chan; Nelson, David; Lautala, Pasi; Jeon, Myounghoon (Elsevier, 2022-09-27)
    Even though the rail industry has made great strides in reducing accidents at crossings, train-vehicle collisions at Highway-Rail Grade Crossings (HRGCs) continue to be a major issue in the US and across the world. In this research, we conducted a driving simulator study (N = 35) to evaluate a hybrid in-vehicle auditory alert (IVAA), composed of both speech and non-speech components, that was selected after two rounds of subjective evaluation studies. Participants drove through a simulated scenario and reacted to HRGCs with and without the IVAA present and through different music conditions and crossing devices. Driver simulator testing results showed that the inclusion of the hybrid IVAA significantly improved driving behavior near HRGCs in terms of gaze behavior, braking reaction, and approach speed to the crossing. The driving simulator study also showed the effects of background music and warning device types on driving performance. The study contributes to the large-scale implementation of IVAAs at HRGCs, as well as the development of guidelines toward a more standardized approach for IVAAs at HRGCs.