32584 research outputs found
Sort by
Sirilla: Predicting Traffic Flow via Stacked Decentralized Federated Learning
Get PDFBillions of people today rely on traffic predictions to optimize their travels. Digital mapping services deliver accurate predictions by learning from vast troves of historical data. Impressive as these systems are, their assumptions do not always apply. They depend on an endless flow of sensitive user data to a central authority, a stable Internet connection, and trustworthiness on both sides of the traditional client-server model. This thesis explores a novel architecture which bucks those assumptions. In the proposed model, traffic data remains on edge devices which individually train models via federated learning. Beyond the obvious privacy benefits, this architecture enables traffic prediction to occur in scenarios which the current paradigm struggles to support. In particular, in disaster scenarios such as earthquakes or hurricanes, evacuees may have no recourse in a dangerous environment without a system such as this
Conscious Control: A Case Study on Self-Monitoring and Interventions in Maladaptive Daydreaming
Get PDFMaladaptive daydreaming (MD) is a persistent mental activity
characterized by immersive, excessive daydreaming that disrupts daily
life. This study examines the effectiveness of self-regulation strategies
in managing MD, including cognitive techniques, behavioral
adjustments, technology-assisted cues, and environmental
modifications. Using a single-subject ABA design, each intervention
phase lasted one week, with MD episodes tracked through self-reported
diaries and EEG monitoring. The results demonstrated that cognitive
reframing and task switching produced the most sustained reductions in
MD episodes, while technology-assisted cues had diminishing
effectiveness over time due to adaptation. Environmental modifications,
particularly workspace reorganization, con- tributed to long-term
improvements, while lighting adjustments had a minimal impact. These
findings suggest that cognitive restructuring, behavioral reinforcement,
and environmental optimization are essential to effectively manage MD.
Future research should explore the long-term sustainability of the
intervention, individual differences in response time, and reinforcement
schedules to improve treatment results
Spartan Daily, August 28, 2025
Get PDFVolume 165, Issue 4https://scholarworks.sjsu.edu/spartan_daily_2025/1046/thumbnail.jp
Spartan Daily, September 4, 2025
Get PDFVolume 165, Issue 6https://scholarworks.sjsu.edu/spartan_daily_2025/1048/thumbnail.jp
Public Access to Federally Funded Research Data: Were the mandates working?
No full textData is the “new oil”, the currency of business in the 21st century and as such research organisations are probing the ways in which their data and the policies behind it are used, reused and shared. In a word, these organisations want to know their data has an impact. The US Federal Government is no different. In 2013, the “Holdren Memo” and the Office of Management and Budget’s M-13-13 laid out principles and goals to make all federally funded data publicly available. These memos applied to agencies with R&D budgets over $100,000,000. Nearly 10 years and many policies later many agencies are now trying to assess the work they have done in this arena. Agencies began implementing their public access plans in 2015 and now, ten years later, many are attempting to identify impacts and quantify the outputs of these policies. The process has brought to light questions such as how do agencies define impact? What do views and downloads mean when applied to datasets? And, are these programs successful? Bringing together strands of research from the author’s in progress PhD thesis (with the Gateway Ph.D. at SJSU), this presentation will cover the policies mandated by the memos, the state of outcomes prior to 2025, and changes in the federal government’s attitude toward public access
Spartan Daily, September 16, 2025
Get PDFVolume 165, Issue 10https://scholarworks.sjsu.edu/spartan_daily_2025/1052/thumbnail.jp
The Second Annual Blockchain Tax Conference on January 24, 2025: Overview of Blockchain Technology and Why It Matters for Tax
Get PDFHuman Contributions to Safety Data Testbed Flight Simulation Study: Data Methods, Processing, and Quality
Get PDFNASA’s System-Wide Safety Project achieved a pivotal milestone through a structured observation simulation experiment aiming to transform safety in aviation. Building on prior initiatives, the study investigates pilot behavior and resilience during challenging scenarios. We studied 24 commercial pilots performing simulated real-world challenges during approach to KCLT such as traffic compression, convective weather, and modulating workload. We acquired standard human factors assessments like NASA-TLX and SART, and also collected a variety of psychophysiological measures such as EEG, ECG, and eye tracking. We employed custom questionnaires and retrospective think-aloud exercises to enable quantifying resilient and safe behavior. Video and audio were also recorded from multiple sources. This paper provides details regarding methodological procedures, data management, and a glimpse at some preliminary analyses. The data and code are publicly available providing a dynamic resource that encourages public contribution to quantification of resilient behavior in commercial airline pilots