A modern approach to data privacy with federated learning

Thumbnail Image
Kalkavan, Ziya Can
Şahinaslan, Ender
Şahinaslan, Önder
Journal Title
Journal ISSN
Volume Title
Maltepe University
Today, information technologies and their usage areas are increasing day by day. Advanced technologies such as the internet of things, smart devices and applications, machine learning and arti cial intelligence are a driving force in the spread of their usage areas. The increase in prevalence and use also increases the production and sharing of data. This increase causes various security problems and concerns in terms of data privacy. Therefore, a balance has to be struck between the need for data sharing and its security. For this purpose, the use of federated learning methods has been examined. Traditional data sharing methods focus on centralized solutions for the processing of private and sensitive data of data subjects, but this causes various problems and raises concerns in the sharing of sensitive data. In the federated learning model, it trains locally without data sharing. It has a distributed arti cial intelligence approach that can run di erent resources together. Thus, it o ers an alternative solution that can help address data privacy concerns arising from the traditional method. In this study, the basic principles, usage areas, advantages and difficulties of federated learning, which is also accepted as a modern approach in data privacy, are discussed. The data and examples obtained in the study will be presented.
Information security , Federated learning , Data privacy , Technology and innovation , Artificial intelligence
Kalkavan, Z. C., Şahinaslan, E., Şahinaslan, Ö. (2023). A modern approach to data privacy with federated learning. Ö. Değer ve H. Çakallı (Ed.), 7th international conference of mathematical sciences ICMS 2023 (72. ss.). Istanbul: Maltepe University.