Reflection on a Productive Sabbatical: From Lab Relocation to Research and Grant Writing

Presented By: Pius Adelani, Ph.D., Associate Professor, Chemistry and Biochemistry

Location: University Center, Conference Room C

Description: During the Spring 2024 semester, I embarked on a sabbatical leave with the primary objective of writing manuscripts, collecting data for ongoing projects, and preparing a grant proposal. Along the way, several additional significant projects emerged, enriching the experience. This presentation highlights the scholarship, mentorship, and administrative initiatives I pursued during my leave, aimed at fostering academic excellence and advancing the mission of St. Mary’s University.

Enhancing Cyber Threat Detection: Adapting Intrusion Detection to Changing Environments

Presented By: Sharmin Aktar, Ph.D., Assistant Professor, Computer Science

Location: University Center, Conference Room C

Description: Network Intrusion Detection Systems (NIDS) are essential for identifying and mitigating cybersecurity threats by monitoring network traffic. However, traditional NIDS models struggle when deployed in new network environments due to domain shifts—distributional differences between the training (source) and deployment (target) domains. These shifts degrade model performance, leading to higher false positive and false negative rates. While domain adaptation techniques have been widely explored in fields like computer vision and natural language processing, their application in NIDS remains relatively underdeveloped.

To address this challenge, we propose a semi-supervised domain adaptation framework for NIDS, designed to enhance model generalization across different network environments. Our approach leverages a small amount of labeled data from the target domain to improve adaptation while maintaining efficiency. Specifically, we integrate Domain-Adversarial Neural Networks (DANN) with a CNN-based feature extractor, enabling the model to learn domain-invariant representations while leveraging limited labeled target data to refine decision boundaries.

We conduct a comprehensive cross-domain evaluation using three dataset pairs: NF-BoT-IoT-v2 vs. NF-ToN-IoT-v2, NFv2-CIC-2018 vs. NFv2-UNSW-NB15, and CIC-IDS2017 vs. CSE-CIC-IDS2018. Our results demonstrate that the proposed framework significantly improves detection accuracy in cross-domain settings compared to fully unsupervised approaches. By balancing adaptability and efficiency, our method provides a practical solution for real-world NIDS deployments, ensuring robust performance despite dynamic network conditions.

Bendford’s Law and the Impact of Direct Blockholders

Presented By: Ying “Caroline” Chen, D.B.A., Assistant Professor of Accounting

Location: University Center, Conference Room C

Description: This study investigates Blockholders’ impact on a firm’s financial reporting through corporate governance. We specifically focus on the effects of Blockholders serving on the board of directors. Using a sample of 7,454 firm-years from 1996 to 2001, we utilize Benford’s Law to determine how director Blockholders impact earnings management associated with rounding reported earnings. We find evidence that director Blockholders lessen this type of earnings management, a trait not shared by other Blockholders. Further evidence suggests that busy boards increase this type of earnings management, implying that our findings may be driven by a lack of monitoring by nondirector Blockholders. We also show that director Blockholders may be able to prevent firms from misrepresenting the value of their inventory. Our paper demonstrates the value of having Blockholders on the board of directors because they can avoid financial misreporting through more robust corporate governance.

Navigating Challenges and Seizing Opportunities: The Transformative Impact of AI on Business Education and Teaching Strategies

Presented By: Denada Ibrushi, Ph.D., Associate Professor, Finance

Location: University Center, Conference Room C

Description: This article explores the transformative impact of artificial intelligence (AI) on business education and teaching strategies, focusing on the instructor’s evolving role as a coach, the use of AI as a student assistant, and the development of AI-related skills. While AI cannot fully replace human instructors—lacking the capacity for existential reflection, value transmission, and societal context—it has significant potential to enhance efficiency, enabling instructors to prioritize student creativity and critical thinking. The study highlights various AI applications, such as ClassPoint AI for interactive quizzes, GradeScope for automated grading, and PowerPoint Speaker Coach for presentation skills, as well as tools like Copilot, Perplexity AI, Quillbot, Canva, Tavily AI, Explain Paper, and Julius AI for data analysis, visualization, AI ethics, and communication skill development. Additionally, the paper emphasizes the importance of aligning AI integration with industry demands by maintaining collaboration with businesses, attending AI workshops and conferences, and organizing campus AI information sessions. These efforts ensure educators effectively incorporate AI to prepare students for a rapidly evolving business environment.

Empowering University Students to Advance Laudato Si’ Goals Through Information and Communication Technologies (ICT)

Presented By: Seong Lim, Ph.D., Professor of Finance and Quantitative Management

Location: University Center, Conference Room C

Description: This study explores how advanced information and communication technologies (ICTs), including AI, blockchain, the Internet, and smart mobile devices, can empower university students to contribute to the goals of Laudato Si’. By integrating these technologies into innovative curricula, student-led research, and community engagement, the paper highlights pathways for fostering sustainability, promoting circular economies, and advancing environmental stewardship. This interdisciplinary approach positions students as catalysts for meaningful action, leveraging digital tools to address ecological and social challenges aligned with the principles of Laudato Si’.

Particle-in-cell coupled with Monte-Carlo Collisions and external circuitry for low-temperature plasma reactors

Presented By: Richard Lombardini, Ph.D., Associate Professor of Physics

Location: University Center, Conference Room C

Description: Low-temperature plasma reactors used for chip fabrication involve complex physical processes that affect the quality of pattern transfer on the semiconductor substrate. Detailed investigations of the near-surface plasma environment require fully-kinetic approaches. We use a particle-in-cell/Monte Carlo collision (PIC/MCC) approach coupled with external circuitry to model realistic operating conditions capture the kinetic effects and measure ion energy and angle distribution functions. We build on the exascale-capable PIC/MCC code, WarpX, and leverage Python interfaces to couple external circuitry.  In this talk, we will demonstrate this inexpensive, non-iterative coupled strategy using a two-dimensional Gaseous Electronics Conference reference cell.

AI for Learning: Lessons from the Classroom at St. Mary’s University

Presented By: Xavier M. Montecel, Ph.D., Assistant Professor of Catholic Theological and Social Ethics

Location: University Center, Conference Room C

Description: To what extent should college teachers guard against student use of AI? To what extent should we endorse and regulate the use of AI to enhance student learning? How can this be done? In this presentation, I share the results of a semester-long experiment in the use of AI for learning with students at St. Mary’s. Participants were registered for two distinct sections of a course in theology. Students in the control section were instructed not to use AI without explicit permission. Students in the experimental section were trained in AI prompting, required to use AI to work toward their learning goals, and asked to report on their interactions with the machine. The results point to several key distinctions that ought to inform AI classroom policy, including product-oriented v. process-oriented use of AI and plagiarism v. self-replacement. Survey data also point to existing AI habits and priorities among students with significant pedagogical implications.

Early Childhood Intervention Services for Neurodiversity and Developmental Differences in African Countries

Presented By: Jillian Pierucci, Ph.D., Professor of Psychology

Location: University Center, Conference Room C

Description: Approximately 12.5% of the world’s children live with a developmental difference (Olusanya et al., 2022) – with a leading neurodevelopmental difference being autism (Scandurra et al., 2019; Aderinto et al., 2023). However, there is a scarcity of services in Africa (Pierucci, 2017; Ruparelia et al., 2016; Newton & Chugani, 2013). This study employed a systematic review of websites to identify available early intervention services in Africa’s 54 countries. Four researchers identified services by entering search terms in three online platforms (i.e., Google, Facebook, LinkedIn). Examples of search terms were as follows: family services, resources for African families, and early childhood intervention (ECI) services. Qualitative and quantitative data from the services’ websites were documented on Google Sheets (e.g., types of institutions, types of services offered, professions and number of service providers). Results revealed that most services are concentrated in urban areas. South Africa has the highest quantity and diversity of resources, while Nigeria and Côte d’Ivoire have fewer. Services are predominantly delivered in-person (84%), with telehealth options limited to training and informational support. More than 90% of families impacted by autism live in a LAMIC (Autism Speaks, 2020), further illustrating the need for African ECI services (Ruparelia et al., 2016). These data are guiding the creation of a virtual map that will be accessible world-wide and lead to a global networking system to make known the available ECI services to optimally scale up these services for children with developmental differences and their families (Franz et al., 2018; Oshodi et al., 2016).

On-the-Go Coupons: Enhancing Marketing Effectiveness Through In-Vehicle Distribution

Presented By: Ajaya Swain, Ph.D., Associate Professor of Quantitative Management

Location: University Center, Conference Room C

Description: Companies continually seek novel strategies to engage their target demographic effectively. Among these strategies, the deployment of in-vehicle coupon distribution has garnered significant attention. This study endeavors to elucidate the fundamental drivers within this framework and evaluate their influence on its efficacy within the contemporary marketing landscape. Utilizing secondary data sources, we employ advanced predictive modeling techniques to identify the key variables impacting in-vehicle coupon distribution. By synthesizing insights derived from these models, we proffer actionable strategies for marketing teams to optimize the efficacy and refinement of this approach. Furthermore, we outline potential avenues for further exploration in this domain.

Prosecutorial Charging and AI: Can AI Solve Issues Related to Discretion, Discrimination, and the Prosecutor Vacancy Crisis?

Presented By: Shem Vinton, J.D., Assistant Professor of Practice of Law

Location: University Center, Conference Room C

Description: Criticisms of prosecutorial charging include the opaqueness of the process and the possible application of racial bias in charging decisions. While these examples are specific, their gravamen is the exercise of prosecutorial discretion. In the last decade, the agendas of many “progressive” elected prosecutors touched on the issue by increasing transparency, declining to prosecute certain crimes, and, in one jurisdiction, using an algorithm to redact race-related information from case narratives.

The recent backlash against progressive prosecutors, termed “rogue” by their critics, included criticism of a common reform policy: the non-enforcement of certain offenses. The central dispute is that prosecutorial discretion does not include the power to decide which laws to enforce. Rather, if a crime is alleged and the available evidence satisfies the elements of the offense to the degree required by law, that crime must be charged and prosecuted. These progressive critics might argue that discretion only comes afterward—in how the case is disposed of.

Jurisprudential conflict aside, this dispute ignores a stark reality. Recent data suggests that major jurisdiction prosecutorial offices often face vacancies ranging from 15% to 33% of their available positions. While seemingly unrelated, this “crisis” compounds issues already blighting the system.

But do all of these issues have a standard solution? Is there some implementation, however radical, that can address not only process and policy issues but also the issue of manpower? This research examines the forgoing issues and suggests employing artificial intelligence to address them.

Enhancing Customer Experiences in Airline Code Sharing: Understanding Impact, Expectations, and Strategies for Improvement

Presented By: Rosa Morales, Digital Marketing major

Research mentor: Zecong Ma, Ph.D.

Location: University Center, Conference Room C

Description: This research project aimed to explore the impact of airline code-sharing on customer satisfaction and identify potential strategies for improving these services. Code-sharing, where multiple airlines share the same flight, has become a common practice in the aviation industry, especially among major alliances like Star Alliance, OneWorld, and SkyTeam. However, the effectiveness of these arrangements in enhancing customer experiences remains a topic of interest. Our study involved surveying 169 participants to gather insights into their familiarity with code-sharing, satisfaction levels, and preferences for potential improvements.