Victimization (V) of Big Data: A Solution Using Federated Learning

The rapid emergence of big data has revolutionized the way organizations perceive and utilize information. With its unparalleled ability to process vast volumes of data at high speeds and handle diverse data types, big data is reshaping industries and enabling evidence-based decision-making. However, the proliferation of big data presents significant privacy challenges. The extensive collection, aggregation, and analysis of diverse datasets can inadvertently expose sensitive personal information, leading to potential breaches of individual privacy. In this work, a new "V" called victimization is introduced as a characteristic of big data. This issue can lead to hazardous consequences. To address the vulnerabilities due to this characteristic, a federated learning approach is proposed as a solution. The proposed approach was tested on two datasets in the domain of health care. The model was also trained using the conventional deep learning approach and Pyspark. The findings in our research suggest that the federated learning approach helps in overcoming those issues leading to victimization without compromising the performance of the model.