Federated learning hides who contributed, not what the model memorizes. New ACL 2026 work: FL models memorize client data - your own prefixes pull your own text back out best - with no clear reduction vs centralized training. Update privacy isn't model privacy. #federatedlearning #privacy #ML

SHADE enables quantum models to forget data without re-measuring quantum states. Using classical shadows recorded during training, federated nodes estimate gradients and apply Newton corrections to remove forget data in <0.2s, matching retraining accuracy.

#QuantumML #FederatedLearning #Research

#KER
A comprehensive survey maps poisoning attacks against federated learning, from Byzantine gradients to backdoor triggers, and benchmarks robust defenses under non-IID data conditions.
Details: https://www.maxapress.com/article/doi/10.48130/ker-0026-0014
#FederatedLearning #PoisoningAttacks

Your local RL round trains on your transcripts alone. Everyone's encrypted updates aggregate into one global - tomorrow's adapter knows work no single user ever did. Your model file is never touched. https://github.com/roger-federated/roger-federated #federatedlearning #selfhosted #FOSS

A new paper proposes Personalized Federated Direct Preference Optimization (pFedDPO), letting each client keep a small local LoRA adapter on a shared base model to better fit heterogeneous…

#FederatedLearning #LoRA #PreferenceOptimization #PrivacyPreservingML
https://arxiv.org/abs/2610.12885

Fed-GRPO proposes a federated extension of Group Relative Policy Optimization for LLM fine-tuning, using GRPO's own reward statistics as zero-cost signals to drive aggregation, local training, and communication.…

#FederatedLearning #LLM #RLHF #PrivacyPreservingML
https://arxiv.org/abs/2610.11502

A new pre-deployment framework estimates task difficulty for federated perception systems by combining intrinsic data properties with client-distribution composition. Experiments on three MNIST variants show strong negative…

#EdgeAI #FederatedLearning #AIResearch
https://arxiv.org/abs/2603.28282

Federated Learning on Stochastic Neural Networks

www.dl.begellhouse.com/journals/558...

#FederatedLearning #NeuralNetworks #MachineLearning

Federated learning's pitch: fine-tune locally, share only tiny LoRA gradients. MineGrad (arXiv 2608.01521) breaks it - a malicious server hides extraction in the pretrained model and reads your training data out of them. The server is the attack surface. #privacy #federatedlearning #ML

This arXiv paper proposes vFedProtoQNAS, a federated learning approach where clients train device-specific quantum neural networks and share class prototypes instead of model parameters. It targets…

#QuantumAI #FederatedLearning #EdgeAI #NeuralArchitectureSearch
https://arxiv.org/abs/2610.01718

Stacking several LoRA deltas onto one base? Summing them scales the effective step with N - each delta was trained to move the weights a given distance, and the sum overshoots every one of them. Average them (1/N) instead. #LoRA #MachineLearning #federatedlearning

A new arXiv study, "Unapologetically Distributed," provides the first broad look at distributed learning in Document Analysis across tasks, architectures, and fine-tuning strategies. It challenges the…

#FederatedLearning #DocumentAnalysis #PrivacyPreserving #AI
https://arxiv.org/abs/2609.39684

New federated learning approach HO-FL splits each model into a zeroth-order bottom segment and a first-order top segment, letting edge devices pick the boundary based on their memory budget while training one…

#FederatedLearning #EdgeAI #ModelTraining #AIResearch
https://arxiv.org/abs/2609.39074

Fixed: roger resolved a fork like gemma-4-12B-it-heretic to its base by name containment - so its chats would have trained the base federation's global. It resolves exactly now. #federatedlearning #opensource #selfhosted https://github.com/roger-federated/roger-federated

Manchester Metropolitan University is offering a PhD Studentship in Lightweight Trust-Aware Private Federated Learning for Secure UAV Swarms under Model Poisoning Attacks.

mlscientist.com/phd-students...

#FederatedLearning

OPFL is a verification framework for privacy-preserving federated learning that replays client training inside secure multi-party computation to detect malicious updates like model…

#FederatedLearning #MachineLearning #PrivacyPreservingAI #SecureComputation
https://arxiv.org/abs/2609.37011

🎉 Paper accepted @vldb.bsky.social 2027

📄 Federated Computation of ROC and PR Curves

📌 TL;DR: Privacy-preserving ROC/PR curve approximation for federated learning

🔗 Project page: xuefeng-xu.github.io/research/fed...

🤝 Joint work with @grahamrc.bsky.social

#VLDB2027 #Privacy #FederatedLearning

A new arXiv paper proposes a federated learning framework for infant movement analysis and General Movement Assessment using skeletal motion data, addressing privacy and governance limits of centralized…

#FederatedLearning #InfantHealth #MedicalAI #ComputerVision
https://arxiv.org/abs/2609.31463

FedHisto-PAST v2 adapts a frozen HIBOU-B pathology encoder for federated lung histopathology classification, handling stain variation and non-IID data across five clients. Internal performance was…

#AI #FederatedLearning #DigitalPathology #Histopathology
https://arxiv.org/abs/2609.31150

🔒 Ready to lock down your IoT edge? Discover how federated learning lets devices train AI while keeping data private. A security game‑changer! #AI #IoT #FederatedLearning #EdgeComputing #Security 🔗