ICML 2026 accepted papers on Federated learning and privacy
130 of the 6,341 papers accepted at ICML 2026 (2.1%) are on Federated learning and privacy, compared with 3.2% at ICML 2025. They fall into 4 topics.
See ICML 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Personalized and heterogeneous FL · 66 papers
- Federated Sketching LoRA: A Flexible Framework for Heterogeneous Collaborative Fine-Tuning of LLMs (1 citation)
- Adaptive Personalized Federated Learning via Multi-task Averaging of Kernel Mean Embeddings
- BESplit: Bias-Compensated Split Federated Learning with Evidential Aggregation
- Decoupled Training with Local Reinforcement Fine-Tuning in Federated Learning
- Distribution Alignment for One-Shot Federated Learning via Optimal Transport
- Federated Distillation for Whole Slide Image via Gaussian-Mixture Feature Alignment and Curriculum Integration
- Federated Learning with Unlabeled Clients: Personalization Can Happen in Low Dimensions
- Federated Variational Preference Alignment with Gumbel-Softmax Prior for Personalized User Preferences
- FedHPro: Federated Hyper-Prototype Learning via Gradient Matching
- FedReLa: Imbalanced Federated Learning via Re-Labeling
- FedRot-LoRA: Mitigating Rotational Misalignment in Federated LoRA
- FedSDR: Federated Self-Distillation with Rectification
and 54 more in the interactive map.
Differential privacy · 40 papers
- Accuracy-First Rényi Differential Privacy and Post-Processing Immunity
- Accurate, Private, Secure, Federated U-statistics with Higher Degree
- ACTG-ARL: Differentially Private Conditional Text Generation with RL-Boosted Control
- Asymptotic Optimality of the High-Dimensional Gaussian Mechanism and Improved Low-Dimensional Mechanisms for Differential Privacy
- Continual Learning With Participation Privacy: An Auditable Buffering-Aggregation Recipe
- Differentially Private Continual Release with Relative Error
- Differentially Private Geodesic Regression
- Differentially Private Preference Data Synthesis for Large Language Model Alignment
- Differentially Private Range Subgraph Counting
- Differentially Private Submodular Maximization with a Knapsack Constraint
- Differentially Private Synthetic Data via APIs 4: Tabular Data
- DP-KFC: Data-Free Preconditioning for Privacy-Preserving Deep Learning
and 28 more in the interactive map.
Privacy attacks and secure learning · 12 papers
- Collaborative Threshold Watermarking
- EPSVec: Efficient and Private Synthetic Data Generation via Dataset Vectors
- On the Fragility of Data Attribution When Learning Is Distributed
- Privasis: Synthesizing the Largest "Public" Private Dataset from Scratch
- $\texttt{FHAIM}$: Fully Homomorphic AIM for Private Tabular Synthetic Data Generation
- An Empirical Study on the Resilience of Partial Merging to Model Clone Attacks
- INDEXGUARD: Index-only Backdoor Vetting for Secure Federated PEFT of Large Language Models
- OptiFluence: Principled Design of Privacy Canaries
- OSNIP: Balancing the Privacy-Utility-Efficiency Trilemma in LLM Inference via Obfuscated Semantic Null Space
- PISA: Privacy-Preserving Split Adaptation with Model IP Protection
- The Hidden Risk: Membership Inference Attacks on Multimodal Federated Learning via Modality Imbalance
- Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data
Communication-efficient FL · 12 papers
- Breaking the Capacity Bottleneck in Model-Heterogeneous Federated Learning via Gradual Model Restoration
- Class-Grouped Normalized Momentum and Faster Hyperparameter Exploration to Tackle Class Imbalance in Federated Learning
- Delayed Momentum Aggregation: Communication-efficient Byzantine-robust Federated Learning with Partial Participation
- Federated Bilevel Performative Prediction
- FedQueue: Queue-Aware Federated Learning for Cross-Facility HPC Training
- HO-SFL: Hybrid-Order Split Federated Learning with Backprop-Free Clients and Dimension-Free Aggregation
- Server-Proximal Aggregation for Federated Domain-Incremental Learning under Partial Participation: Task-Uniform Convergence and Backward Transfer
- DecFus: Decentralized Layer-wise Fusion with Dynamic Exploration and Exploitation
- FedFit: Federated Dynamic Sparse Training via Fisher Information scoring
- No Retraining at Edge: Efficient Resource-Aware Mixed-Precision Quantization via Federated Supernet Learning
- Robust Federated Learning Against Adaptive Compression
- ShapCCS: Shapley-Driven Client Coreset Selection in Federated Learning