# Privacy-Aligned Personalized Federated Learning with Compact Adaptation and Variable-Length Gaussian Communication

arXiv: https://arxiv.org/abs/2609.15950 · HTML: https://mlsanity.com/paper/2609.15950/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 18 references that are in the map.
- First posted on arXiv 14 Sep 2026.

## Builds on (most cited first, 9 of 9)

- [Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification](https://mlsanity.com/paper/1909.06335/) (2019)
- [Personalized Federated Learning using Hypernetworks](https://mlsanity.com/paper/2103.04628/) (2021)
- [Large Scale Private Learning via Low-rank Reparametrization](https://mlsanity.com/paper/2106.09352/) (2021)
- [PeFLL: Personalized Federated Learning by Learning to Learn](https://mlsanity.com/paper/2306.05515/) (2023)
- [Compression with Exact Error Distribution for Federated Learning](https://mlsanity.com/paper/2310.20682/) (2023)
- [Communication Efficient Private Federated Learning Using Dithering](https://mlsanity.com/paper/2309.07809/) (2023)
- [Rejection-Sampled Universal Quantization for Smaller Quantization Errors](https://mlsanity.com/paper/2402.03030/) (2024)
- [Deploying Models to Non-participating Clients in Federated Learning without Fine-tuning: A Hypernetwork-based Approach](https://mlsanity.com/paper/2508.12673/) (2025)
- [Private Model Personalization Revisited](https://mlsanity.com/paper/2506.19220/) (2025)

## Abstract (arXiv)

Record-level differential privacy exposes a structural misalignment in personalized federated learning when client-specific variation is low-dimensional while training repeatedly releases high-dimensional updates. In this paper, we address this misalignment by releasing a private client context once and confining repeated adaptation to a fixed coefficient space. Beyond dimensionality reduction, the factorized generator induces an adaptive optimization geometry that reshapes noisy updates, and controlled ablations show that most of its private-training gain is retained by radial evolution. To further reduce the communication cost, we realize the Gaussian mechanism for coefficient updates directly through variable-length quantization with finite expected code length, so that the quantization error itself serves as the required privacy perturbation rather than extra distortion. Across MNIST and CIFAR-10, our design matches or outperforms full-model private adaptation across privacy budgets and client heterogeneity, while reducing protected uplink by a factor of 2.67 at (varepsilon=16) on CIFAR-10 with comparable future-client accuracy.