privacy-preserving

This paper proposes a Federated Hash Projected Latent Factor (FHPLF) model that integrates hash learning into federated learning to reduce communication costs and enhance privacy, using binary gradient-like matrices and projected Hamming distance to improve accuracy and efficiency.

This paper systematically evaluates three survival models (Cox, DeepSurv, RSF) under federated learning on heterogeneous breast cancer data, finding that FL outperforms local training and RSF offers the best balance of performance across clients.

This survey provides a systematic review of federated causal discovery and inference, organizing methods by methodological paradigm, federation topology, and structural scope, and highlighting open challenges.

PSyGenTAB is a privacy-preserving framework that uses constrained optimization to generate synthetic clinical tabular data, balancing privacy and utility while preserving clinical relationships and minority-class patterns.

This paper introduces DiSan, a privacy-preserving text sanitization framework for distributed agent collaboration. By disentangling source-invariant role content from source-identifying style, DiSan reduces PII exposure 20× while maintaining 83% answer faithfulness on a multi-agent RAG benchmark, outperforming traditional masking approaches.

MedLatentDx proposes a latent multi-agent communication framework for cross-hospital rare-disease diagnosis, using latent KV blocks to share diagnostic evidence without exposing clinical text, and introduces the CrossRare-Bench benchmark.

This paper provides a comprehensive survey of Federated Continual Learning (FCL), an emerging field that combines Federated Learning and Continual Learning to enable lifelong, adaptive, and privacy-preserving learning over distributed and non-stationary data. It proposes a taxonomy, reviews applications, metrics, and open challenges.

This paper presents a vision for integrating multi-modal multi-task federated foundation models (M3T FedFMs) into vehicular networks, discussing training principles, use cases, challenges, and a case study on the Waymo Open Dataset.

InfoShield introduces a privacy-preserving method for speech representations in mental health screening using information-theoretic optimization, reducing sensitive attribute inference while maintaining diagnostic accuracy. A novel TimeAwareMINE estimator addresses temporal-static misalignment in sequential speech.

This paper proposes a locally deployed RAG-based academic advising system that combines large language models with retrieval from structured syllabus data to support course selection and personalized study planning in a privacy-preserving manner.

LLM-FACETS is an open-source evaluation framework designed to help practitioners assess LLM transparency and accountability with a focus on privacy and data flow transparency. It provides a browser interface, plugin architecture, and supports multiple auditing mechanisms including token-level log-probability visualization and RAG Triad metrics.

This paper proposes CE-FedGNN, a federated graph neural network framework that achieves communication efficiency and privacy preservation by infrequently exchanging aggregated node representations with metric differential privacy guarantees, and demonstrates strong performance on benchmarks.

The Chromium team proposes a new Embedding API for the web platform that allows developers to generate vector embeddings on-device using Chrome's AI infrastructure, enabling privacy-preserving semantic search, retrieval-augmented generation, and content clustering while reducing latency and cost.

PrivFusion is a privacy-preserving multi-agent framework that automates the harmonization of structured datasets across institutions before federated training, reducing manual effort and enabling collaborative analytics on sensitive clinical data.

This paper presents FederatedRSF, a Python package for federated random survival forests that handles partially overlapping medical data across institutions without sharing raw data, and demonstrates comparable performance to centralized training on breast cancer data.

This paper presents a privacy-preserving federated recommendation system for mobile devices, using a two-stage pipeline with candidate generation and ranking, implemented via Kotlin Multiplatform on Android/iOS.

This paper introduces FIRMA, a family of three privacy-preserving federated learning protocols using Fibonacci-weighted ring aggregation to achieve server-free operation, permanently private classification heads, and improved accuracy under data heterogeneity.

Proposes M²FedAQI, a lightweight multimodal federated learning framework for air quality prediction across heterogeneous edge devices, achieving significant improvements over baselines on benchmark datasets.

This paper introduces AgentStop, a lightweight supervisor that predicts and preemptively terminates local AI agent trajectories unlikely to succeed, reducing energy waste by 15-20% with minimal impact on task performance.

This paper demonstrates that small open-weight LLMs (<30B parameters) can achieve competitive interpretable translation quality estimation, including MQM error annotations and corrections, rivaling much larger proprietary models while preserving data privacy.