April 4, 2023
The article titled “Privacy-preserving Federated Learning and its application to natural language processing” written by Balázs Nagy, István Hegedűs, Noémi Sándor, Balázs Egedi, Haaris Mehmood, Karthikeyan Saravanan, Gábor Lóki, and Ákos Kiss has been accepted into the prestigious journal Knowledge-Based Systems.
The researchers were trying to provide a solution for the privacy concerns of edge devices, namely issues like sending sensitive data to online servers. They stated that while Federated Learning (FL) approaches can help in these situations, they alone cannot solve all challenges. Therefore, they proposed a privacy-preserving FL framework, which leverages the concepts of bitwise quantization, local differential privacy (LDP), and feature hashing for input representation in the collaborative training of ML models.
They demonstrated that their approach is a feasible solution for private language processing tasks on edge devices without the use of resource-hungry language models or privacy-violating collection of client data.
The full paper can be read on ScienceDirect.