Federated machine learning and differential privacy in big data processing: a curriculum module for pre-service informatics teacher education
The centralized model of big-data processing—collecting learner data into a single repository to train predictive models—increasingly conflicts with data-protection regulation and with the institutional fragmentation of educational data. Pre-service informatics teacher education programs, however, still center on centralized big-data architectures (Hadoop, Spark, relational and NoSQL databases), largely omitting the decentralized, privacy-oriented approaches that have gained prominence in the recent research literature. This article presents a curriculum module, “Federated Machine Learning and Differential Privacy in Educational Data,” designed for integration into pre-service informatics teacher education programs, together with a description of its initial quasi-experimental implementation. The module's pedagogical framework draws on constructivist learning theory, a Backward Design course-development model, and Universal Design for Learning (UDL) principles for accommodating diverse learners, and is delivered across two sub-modules comprising 15 lectures and 30 practical sessions built around a practice-oriented project using students’ own programming-course data. The module was piloted with 119 pre-service informatics teachers across two Kazakhstani universities: an experimental group ( n = 94) drawn from three cohorts (2022–2023, 2023–2024, and 2024–2025) enrolled in the “6B01511—Informatics” program at L.N. Gumilyov Eurasian National University, and a nonequivalent control group ( n = 25) enrolled in the “6B01503—Informatics Teacher Training” program at M.Kh. Dulaty Taraz University. We describe the assessment instruments—competency rubrics, case-analysis tasks, and a pre/post confidence survey—used to evaluate technical-technological, legal-ethical, and pedagogical competencies, together with the sample composition and implementation process; full outcome analysis comparing the experimental and control groups is in progress and will be reported in a follow-up publication. We discuss practical implications for institutions considering adoption of the module and explicitly acknowledge conceptual, methodological (including the nonrandomized, institution-level assignment of groups), environmental, and material constraints of the current implementation. Article type: this is a curriculum-design and implementation report. It does not report comparative or effectiveness outcomes; given the non-randomized, institution-level group assignment and the small, unequal control group ( n = 25 vs. n = 94), no causal or comparative claims are made or intended in this article.
Authors
- Gulmira Yerlanova (ORCID: https://orcid.org/0000-0002-2817-7110)
- Меруерт Серік (ORCID: https://orcid.org/0000-0002-2801-432X)
- Г.Ф. Нурбекова (ORCID: https://orcid.org/0009-0005-8468-9523)
- Madina Maulenova
Institutions
- L. N. Gumilyov Eurasian National University (KZ)
- Semey Medical University (KZ)
- Alikhan Bokeikhan University (KZ)
Publication Details
- Journal
- Frontiers in Education
- Published
- 2026-11-18
- DOI
- https://doi.org/10.3389/feduc.2026.1941774
- Primary Topic
- Teaching and Learning Programming
- Type
- article
- Field-Weighted Citation Impact
- 0.00