arXiv:2605.30075v1 Announce Type: new 
Abstract: Quantum Federated Learning (QFL) offers a promising framework to train quantum models across distributed clients while keeping data strictly local. Due to its simplicity and low communication overhead, Federated Averaging (FedAvg) is the standard aggregation choice in QFL literature. However, deploying QFL on practical hardware exposes a severe double-drift phenomenon: the global model is simultaneously derailed by client drift from non-IID data and hardware bias from noisy quantum gradient estimates. In this work, we first analyze the convergence of FedAvg under these realistic conditions, mathematically demonstrating that quantum hardware bias creates a persistent error floor that standard averaging cannot correct. To overcome this limitation, we propose Q-ANCHOR, a quantum-aware federated aggregation architecture that anchors server updates with zero-noise extrapolation while applying stateful client correction to suppress both client drift and hardware-induced bias. Our convergence theory proves that Q-ANCHOR successfully mitigates classical client drift while actively reducing the hardware-bias floor. Experimental results demonstrate that Q-ANCHOR achieves significantly more stable training than conventional FL baselines.

تم تقديم إطار عمل جديد يسمى Q-ANCHOR لتحسين التعلم الفيدرالي الكمي (QFL) من خلال معالجة ظاهرة الانحراف المزدوج التي تؤثر على تدريب النماذج على الأجهزة العملية. يستخدم هذا الإطار الاستقراء بدون ضوضاء لتثبيت تحديثات الخادم بينما ينفذ تصحيحًا لحالة العميل لتخفيف الأخطاء الناتجة عن البيانات غير المستقلة وغير المتطابقة (non-IID) والتحيزات الناتجة عن الأجهزة.

Se ha introducido un nuevo marco llamado Q-ANCHOR para mejorar el Aprendizaje Federado Cuántico (QFL) al abordar el fenómeno de doble deriva que afecta el entrenamiento de modelos en hardware práctico. Este marco utiliza la extrapolación sin ruido para anclar las actualizaciones del servidor mientras implementa una corrección de cliente con estado para mitigar los errores causados por datos no IID y sesgos de hardware.

Un nouveau cadre appelé Q-ANCHOR a été introduit pour améliorer l'apprentissage fédéré quantique (QFL) en abordant le phénomène de double dérive qui affecte l'entraînement des modèles sur du matériel pratique. Ce cadre utilise l'extrapolation sans bruit pour ancrer les mises à jour du serveur tout en mettant en œuvre une correction client d'état pour atténuer les erreurs causées par des données non IID et un biais matériel.

A new framework called Q-ANCHOR has been introduced to enhance Quantum Federated Learning (QFL) by addressing the double-drift phenomenon that affects model training on practical hardware. This framework utilizes zero-noise extrapolation to anchor server updates while implementing stateful client correction to mitigate errors caused by non-IID data and hardware bias.

Q-ANCHOR: Federated Quantum Learning with ZNE-guided Correction

arXiv:2604.24012v3 Announce Type: replace 
Abstract: Federated learning enables a population of clients to collaboratively train machine learning models without exchanging their raw data, but standard algorithms such as FedAvg suffer from slow convergence and high communication and memory costs in heterogeneous, resource-constrained environments. We introduce FedSLoP, a federated optimization algorithm that combines stochastic low-rank subspace projections of gradients, thereby reducing the dimension of communicated and stored updates while preserving optimization progress. On the theoretical side, we develop a detailed nonconvex convergence analysis under standard smoothness and bounded-variance assumptions, showing that FedSLoP is guaranteed to converge to a first-order stationary point at a rate of $O(1/\sqrt{NT})$. On the empirical side, we conduct extensive experiments on federated MNIST classification with heterogeneous data partitions, showing that FedSLoP substantially reduces communication volume and client-side memory while achieving competitive or better accuracy compared with FedAvg and representative sparse or low-rank baselines. Together, our results demonstrate that random subspace momentum methods such as FedSLoP provide a principled and effective approach to communication- and memory-efficient federated learning. Codes are available at: https://github.com/pkumelon/FedSLoP.git.

تم تقديم FedSLoP، وهو خوارزمية تعلم فدرالي فعالة من حيث الذاكرة، لمعالجة التحديات المتعلقة بالتحول البطيء وارتفاع تكاليف الاتصال المرتبطة بالطرق التقليدية مثل FedAvg في البيئات غير المتجانسة. من خلال استخدام إسقاطات فرعية عشوائية لمتجهات التدرج، تقلل FedSLoP من أبعاد التحديثات مع ضمان تقدم التحسين.

La introducción de FedSLoP, un algoritmo de aprendizaje federado eficiente en memoria, aborda los desafíos de la lenta convergencia y los altos costos de comunicación asociados con métodos tradicionales como FedAvg en entornos heterogéneos. Al utilizar proyecciones estocásticas de subespacios de gradientes, FedSLoP reduce la dimensionalidad de las actualizaciones mientras garantiza el progreso de la optimización.

L'introduction de FedSLoP, un algorithme d'apprentissage fédéré économe en mémoire, répond aux défis de la lente convergence et des coûts de communication élevés associés aux méthodes traditionnelles comme FedAvg dans des environnements hétérogènes. En utilisant des projections stochastiques de sous-espaces de gradients, FedSLoP réduit la dimensionnalité des mises à jour tout en garantissant le progrès de l'optimisation.

The introduction of FedSLoP, a memory-efficient federated learning algorithm, addresses the challenges of slow convergence and high communication costs associated with traditional methods like FedAvg in heterogeneous environments. By utilizing stochastic low-rank subspace projections of gradients, FedSLoP reduces the dimensionality of updates while ensuring optimization progress.

Q-ANCHOR: Federated Quantum Learning with ZNE-guided Correction

Was this article worth reading? Share it

Ready to build your own newsroom?