Content uploaded by Mana-Teresa Donner
Author content
All content in this area was uploaded by Mana-Teresa Donner on Mar 12, 2025
Content may be subject to copyright.
A.1 Practical Applications of Fed_LLM for Privacy-
Preserving AI in Education
Sandra Hummel1, Gitanjali Wadhwa1, Syed Hur Abbas1,
Mana-Teresa Donner1
1 TUD Dresden University of Technology, ScaDS.AI Dresden
1 Introduction
The Fed_LLM framework integrates federated learning and large language models
(LLMs) to create a cutting-edge educational system. This innovative approach
optimizes learning experiences and ensures robust data privacy, making it an
essential tool for modern education. Fed_LLM is practically applied and tested in
educational settings, focusing on its products, concepts, and real-world
implementation.
In educational settings, the demand for personalized, efficient, and secure learning
tools is at an all-time high (Hummel et al., 2023; Hummel & Donner, 2023; Bartok
et al., 2023). Traditional educational systems often struggle to meet these needs
(Egger & Hummel, 2020), especially when it comes to maintaining data privacy
while providing tailored learning experiences. The Fed_LLM framework addresses
these challenges by combining the decentralized nature of federated learning with
the advanced capabilities of LLMs. This combination not only personalizes
educational content to cater to individual student needs but also ensures that
sensitive data remains protected on local devices (Li et al., 2020; Yang et al., 2019).
The impressive performance of generative pre-trained LLMs and their multi-modal
derivations largely owes to their capacity to learn representations at scale. However,
this capability comes with significant energy and monetary costs due to the
extensive computational resources required for training these models (Hoffmann et
al., 2022). Traditionally, training large LLMs involves thousands of hours of
computation on specialized hardware in centralized data centers, a process that
incurs high energy and monetary costs. Distributing training across multiple data
centers can further escalate these costs due to communication overheads (Lin et al.,
2020). Federated learning (FL) presents a promising solution by enabling
8
professional development programs should be established to assist educators in
integrating Fed_LLM into their pedagogy.
Although federated learning addresses many privacy concerns by keeping data
localized, continuous monitoring and robust cybersecurity measures are necessary
to protect student data from potential breaches. The integration of Fed_LLM into
educational practice must be pedagogically sound, aligning with educational goals
and enhancing the learning process. Research should explore the most effective
ways to integrate Fed_LLM into various educational settings to maximize its
benefits. Implementing Fed_LLM involves significant financial and resource
investments, so educational institutions must carefully consider their budgets and
allocate resources effectively to ensure the sustainability of the technology.
Partnerships with technology companies and government funding could help
alleviate some financial burdens.
The ethical implications of using AI in education must also be thoroughly examined,
including addressing potential biases in AI algorithms and ensuring that the
technology promotes fairness and inclusivity. By addressing these critical
perspectives and leveraging the innovative capabilities of Fed_LLM, educational
institutions can significantly enhance the quality of education, providing a more
personalized, efficient, and secure learning environment for all students. This
comprehensive approach ensures that the transformative potential of Fed_LLM is
fully realized, marking a significant step forward in the advancement of educational
technology.
References
Bartok, L., Donner, M.-T., Ebner, M., Gosch, N., Handle-Pfeiffer, D., Hummel, S.,
Kriegler-Kastelic, G., Leitner, P., Tang, T., Veljanova, H., Winter, C., &
Zwiauer, C. (2023). Learning Analytics – Studierende im Fokus. Zeitschrift
für Hochschulentwicklung: ZFHE; Beiträge zu Studium, Wissenschaft und
Beruf, 18 (Sonderheft Hochschullehre), 223-250.
https://doi.org/10.3217/zfhe-SH-HL/12
Bohlinger, S., & Hummel, S. (2024). Digital Capacity Building in Teacher
Education: An Environmental Case Study from Cambodia. In S. Hummel
(Ed.), Empowering Education in Cambodia and Sri Lanka: Advancing
9
Quality in 21st Century Teaching and Learning (pp. 31-50). Springer.
https://doi.org/10.1007/978-3-658-43306-2_3
Carlini, N., Liu, C., Erlingsson, U., Kos, J., & Song, D. (2019). The Secret Sharer:
Evaluating and Testing Unintended Memorization in Neural Networks. In
28th USENIX Security Symposium (USENIX Security 19) (pp. 267-284).
Santa Clara, CA: USENIX Association.
https://www.usenix.org/conference/usenixsecurity19/presentation/carlini
Egger, R., & Hummel, S. (2020). Stolperstein oder Kompetenzstufe. Die
Studieneingangsphase und ihre Bedeutung für die Wissenschaftssozialisation
von Studierenden. Springer VS. https://doi.org/10.1007/978-3-658-23283-2
Geyer, R. C., Klein, T., & Nabi, M. (2017). Differentially Private Federated
Learning: A Client-Level Perspective. arXiv preprint arXiv:1712.07557.
https://arxiv.org/abs/1712.07557
Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E.,
Millican, K., van den Driessche, G., Lespiau, J. B., Bennett, L., Schmid, L.,
Scicluna, M., Zimermann, T., Rae, J. W., Veličković, P., Ring, R.,
Babuschkin, I., Martens, L., ... Irving, G. (2022). Training Compute-Optimal
Large Language Models. arXiv preprint arXiv:2203.15556.
https://arxiv.org/abs/2203.15556
Hummel, S., Donner, M.-T., & Egger, R. (2023). Lehren mit Learning Analytics.
Qualifizierungsprogramm für die Hochschullehre. Universität Graz.
https://oer-portal.uni-graz.at/edu-
sharing/components/collections?viewType=1&id=df2fc7c7-b832-421b-
acec-764181c6c2f0
Hummel, S. (2020). Grundlagen der Hochschullehre. Springer.
https://doi.org/10.1007/978-3-658-28181-6
Hummel, S., Egger, R., & Donner, M.-T. (2023). Lehren mit KI: Learning Analytics
für mehr Studierendenorientierung in der Hochschullehre? In T. Köhler, E.
Schopp, N. Kahnwald & R. Sonntag (Eds.), Communities in New Media.
Inclusive digital: Forming Community in an Open Way Self-determined
Participation in the Digital Transformation: Proceedings of 26th Conference
GeNeMe (pp. 7-16). TUDpress. https://doi.org/10.25368/2024.7
Hummel, S., & Donner, M.-T. (2023). KI-Anwendungen in der Hochschulbildung
aus Studierendenperspektive. FNMA Magazin, 3, 38-41.
https://doi.org/10.13140/RG.2.2.13967.97446
Hummel, S., Aldrian, S., & Sheehan, B. (2024). Culture, Education and Digital
Transformation in Sri Lanka. In S. Hummel (Ed.), Empowering Education in
Cambodia and Sri Lanka: Advancing Quality in 21st Century Teaching and
Learning (pp. 9-29). Springer. https://doi.org/10.1007/978-3-658-43306-2_2
Hummel, S., Brodacz-Geier, M., & Sheehan, B. (2024). Pedagogical
Professionalization in Digital Contexts. In S. Hummel (Ed.), Empowering
10
Education in Cambodia and Sri Lanka: Advancing Quality in 21st Century
Teaching and Learning (pp. 203-219). Springer. https://doi.org/10.1007/978-
3-658-43306-2_13
Hummel, S., Bohlinger, S., & Sheehan, B. (2024). Modern Pedagogies in the Digital
Era: Elevating Educational Standards in Teacher Education. In S. Hummel
(Ed.), Empowering Education in Cambodia and Sri Lanka: Advancing
Quality in 21st Century Teaching and Learning (pp. 65-76). Springer.
https://doi.org/10.1007/978-3-658-43306-2_5
Kairouz, P., McMahan, H. B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A. N., ...
& Zhao, S. (2021). Advances and Open Problems in Federated Learning.
Foundations and Trends® in Machine Learning, 14(1-2), 1-210.
https://doi.org/10.1561/2200000083
Li, T., Sahu, A. K., Talwalkar, A., & Smith, V. (2020). Federated Learning:
Challenges, Methods, and Future Directions. IEEE Signal Processing
Magazine, 37(3), 50-60. https://doi.org/10.1109/MSP.2020.2975749
McMahan, H. B., Moore, E., Ramage, D., Hampson, S., & y Arcas, B. A. (2017).
Communication-Efficient Learning of Deep Networks from Decentralized
Data. In Artificial Intelligence and Statistics (pp. 1273-1282). PMLR.
https://arxiv.org/abs/1602.05629
Rieke, N., Hancox, J., Li, W., Milletari, F., Roth, H. R., Albarqouni, S., ... & Bakas,
S. (2020). The Future of Digital Health with Federated Learning. NPJ Digital
Medicine, 3(1), 1-7. https://doi.org/10.1038/s41746-020-00323-1
Yang, Q., Liu, Y., Chen, T., & Tong, Y. (2019). Federated Machine Learning:
Concept and Applications. ACM Transactions on Intelligent Systems and
Technology (TIST), 10(2), 1-19. https://doi.org/10.1145/3298981
