Physics-Informed Pedagogy in AI Engineering: A Case Study on Teaching Fundamental Algorithms via Signal Processing

Носенко, Тетяна Іванівна (2026) Physics-Informed Pedagogy in AI Engineering: A Case Study on Teaching Fundamental Algorithms via Signal Processing Computer Applications in Engineering Education, 3 (34). pp. 1-10. ISSN 1099-0542

Text NosenkoPIPAIIE.pdf Download (1MB)

Abstract

The rapid integration of artificial intelligence tools into engineering education creates the risk of a “black-box” approach, where students utilize high-level ML libraries without a deep understanding of fundamental algorithms. This study proposes and empirically validates the physics-informed scaffolding methodology, which leverages stochastic physical simulations as structured pedagogical support for teaching basic algorithms in introductory programming courses. A case study (N = 34) focused on localizing radiation anomalies in noisy data was conducted. Students implemented fundamental signal processing algorithms (smoothing and thresholding) from scratch, replicating the micro-logic of mathematical convolution. This task was framed as a fundamental signal processing problem, providing a bridge between basic algorithmic structures and AI engineering applications. The effectiveness of the approach was measured using a validated instrument assessing three constructs: Engagement, Conceptual Understanding, and Scaffolding and Process Support (SPS). Quantitative analysis confirmed the high reliability of the instrument (Cronbach's α = 0.924). An ordinary least squares regression model demonstrated that the proposed pedagogical support and student engagement are statistically significant predictors, explaining over 70% of the variance (R² = 0.718, p < 0.001) in students' perceived conceptual understanding of the algorithms. The research empirically demonstrates the efficacy of transitioning from a traditional “Syntax-First” to a “Data-First” educational model. Engaging with real-world physical noise helps students overcome “stochastic shock” and fosters the engineering intuition necessary for effective problem-solving in AI engineering. © 2026 Wiley Periodicals LLC.

Item Type:	Article
Uncontrolled Keywords:	algorithmic thinking; CS1 education; engineering education; instructional scaffolding; physics-informed machine learning (PIML)
Subjects:	Статті у базах даних > Scopus > У виданнях Q1 Scopus
Divisions:	Факультет інформаційних технологій та математики > Кафедра комп'ютерних наук
Depositing User:	доцент Тетяна Іванівна Носенко
Date Deposited:	04 Jun 2026 13:52
Last Modified:	04 Jun 2026 13:52
URI:	https://elibrary.kubg.edu.ua/id/eprint/57894

Actions (login required)

View Item