FreqX: Analyze the Attribution Methods in Another Domain
This work addresses interpretability needs in PFL for clients seeking privacy and efficiency, but it appears incremental as it builds on existing interpretability methods with new domain-specific applications.
The paper tackles the challenges of Non-IID data, heterogeneity, fairness, and unclear contributions in Personalized Federated Learning (PFL) by proposing FreqX, a novel interpretability method that uses Signal Processing and Information Theory to provide both attribution and concept information, achieving at least 10 times faster runtime than baselines.
Personalized Federal learning(PFL) allows clients to cooperatively train a personalized model without disclosing their private dataset. However, PFL suffers from Non-IID, heterogeneous devices, lack of fairness, and unclear contribution which urgently need the interpretability of deep learning model to overcome these challenges. These challenges proposed new demands for interpretability. Low cost, privacy, and detailed information. There is no current interpretability method satisfying them. In this paper, we propose a novel interpretability method \emph{FreqX} by introducing Signal Processing and Information Theory. Our experiments show that the explanation results of FreqX contain both attribution information and concept information. FreqX runs at least 10 times faster than the baselines which contain concept information.