FS-TFP/materials/paper_list/Cross-device & PFL/README.md

Cross-Device and Personalized Federated Learning

This list is constantly being updated. Feel free to contribute!

2023

Title	Venue	Link	Keywords	Note
Towards Real-World Cross-Device Federated Learning	KDD	pdf	system, cross-Device, heterogeneous device-runtime, FS-Real
Personalized Federated Learning with Parameter Propagation	KDD	pdf	adaptive parameter propagation, selective regularization, FEDORA
FedCP: Separating Feature Information for Personalized Federated Learning via Conditional Policy	KDD	pdf	conditional policy, partial model parameters
Efficient Personalized Federated Learning via Sparse Model-Adaptation	ICML	pdf	sparse models, conditional policy, cross-Device, pFedGate
Personalized Federated Learning with Inferred Collaboration Graphs	ICML	pdf	collaboration graph, pFedGraph, poisoning attacks
DoCoFL: Downlink Compression for Cross-Device Federated Learning	ICML	pdf	communication compression, cross-device, anchor
Personalized Federated Learning with Feature Alignment and Classifier Collaboration	ICLR	pdf	Collaboration	feature alignment by regularization, theoretically-guaranteed heads combination
Test-Time Robust Personalization for Federated Learning	ICLR	pdf	Test-time Robustness
A Statistical Framework for Personalized Federated Learning and Estimation: Theory, Algorithms, and Privacy	ICLR	pdf	Statistical Estimation, Differential Privacy, Empirical/Hierarchical Bayes
FiT: Parameter Efficient Few-shot Transfer Learning for Personalized and Federated Image Classification	ICLR	pdf	few-shot learning, transfer learning
PerFedMask: Personalized Federated Learning with Optimized Masking Vectors	ICLR	pdf	Masking vectors
The Best of Both Worlds: Accurate Global and Personalized Models through Federated Learning with Data-Free Hyper-Knowledge Distillation	ICLR	pdf	Knowledge Distillation, Differential Privacy,	share means of local data representations and soft predictions; no public data

2022

Title	Venue	Link	Keywords	Note
FedPop: A Bayesian Approach for Personalised Federated Learning	NeurIPS	pdf	Uncertainty quantification; Markov Chain Monte Carlo (MCMC)
pFL-Bench: A Comprehensive Benchmark for Personalized Federated Learning	NeurIPS	pdf, code	Benchmark	10+ dataset variants, 20+ methods, fruitful metrics and settings.
Self-Aware Personalized Federated Learning	NeurIPS	pdf	Uncertainty quantification; Bayesian hierarchical models
Factorized-FL: Personalized Federated Learning with Parameter Factorization & Similarity Matching	NeurIPS	pdf	Kernel Factorization
On Sample Optimality in Personalized Collaborative and Federated Learning	NeurIPS	pdf	Information theoretic bounds; Sample Complexity
Personalized Federated Learning towards Communication Efficiency, Robustness and Fairness	NeurIPS	pdf	Infimal convolution; Low-dimensional projection
DisPFL: Towards Communication-Efficient Personalized Federated Learning via Decentralized Sparse Training	ICML	pdf	Decentralized FL; Sparse Models
Personalization Improves Privacy-Accuracy Tradeoffs in Federated Learning	ICML	pdf	User-level DP; (\epsilon, $delta$)-DP
Personalized Federated Learning through Local Memorization	ICML	pdf	Model interpolation; kNN;
Personalized Federated Learning via Variational Bayesian Inference	ICML	pdf	Bayesian variational inference; Upper bound
Federated Learning with Partial Model Personalization	ICML	pdf	Partial model parameters; Transformer
On Bridging Generic and Personalized Federated Learning for Image Classification	ICLR	pdf	Partial model parameters;
FedBABU: Toward Enhanced Representation for Federated Image Classification	ICLR	pdf	Partial model parameters;	keep the head (classifer) unchanged during FL training, then conduct fine-tuning before inference
Towards Personalized Federated Learning	Transactions on Neural Networks and Learning Systems	pdf	Survey

2021

Title	Venue	Link	Keywords	Note
Federated muli-task learning under a mixture of distributions	NeurIPS	pdf, code	Distribution Mixture; Expectation-Maximization; FedEM
Parameterized Knowledge Transfer for Personalized Federated Learning	NeurIPS	pdf	Knowledge Distillation	transmit only soft-predictions; public dataset required
Personalized Federated Learning with Gaussian Processes	NeurIPS	pdf, code	Gaussian process; Generalization bound
Ditto: Fair and robust federated learning through personalization	ICML	pdf, code	Threat model; Fairness; Regularizer
Personalized Federated Learning using Hypernetworks	ICML	pdf, code	Hypernetwork; Client Embedding
Exploiting Shared Representations for Personalized Federated Learning	ICML	pdf, code	Partial model parameters	FedRep, shared body (feature extractor), personalized head (classifier)
Personalized Federated Learning with First Order Model Optimization	ICLR	pdf, code	Model mixture
FedBN: Federated Learning on Non-IID Features via Local Batch Normalization	ICLR	pdf, code	Partial model parameters

2020

Title	Venue	Link	Keywords	Note
Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach	NeurIPS	pdf	MAML; Per-FedAvg
Personalized federated learning with moreau envelopes	NeurIPS	pdf, code	pFedMe; Regularizer;
An efficient framework for clustered federated learning	NeurIPS	pdf, code	Iterative clustering; IFCA
Adaptive personalized federated learning	arXiv	pdf, code	Model Mixture; APFL
Lower bounds and optimal algorithms for personalized federated learning	NeurIPS	pdf	Communication complexity;
Personalized Federated Learning With Differential Privacy	IEEE Internet of Things Journal	pdf	(\epsilon, $delta$)-DP
Personalized federated learning for intelligent IoT applications: A cloud-edge based framework	IEEE Open Journal of the Computer Society	pdf	Device Heterogeneity
Survey of Personalization Techniques for Federated Learning	2020 Fourth World Conference on Smart Trends in Systems, Security and Sustainability (WorldS4)	pdf	Survey	4 pages, 7 types of personalization methods.

2019

Title	Venue	Link	Keywords	Note
Federated Evaluation of On-device Personalization	arXiv	pdf	Personalized Hyper-parameters; Next-word prediction; RNN	1. Evaluation scale: tens of millions of users. 2. Evaluation method: testing both global and local models on local test set, calculating and uploading the accuracy delta.