Filters








7 Hits in 24.3 sec

FjORD: Fair and Accurate Federated Learning under heterogeneous targets with Ordered Dropout [article]

Samuel Horvath, Stefanos Laskaridis, Mario Almeida, Ilias Leontiadis, Stylianos I. Venieris, Nicholas D. Lane
2022 arXiv   pre-print
We further show that for linear maps our Ordered Dropout is equivalent to SVD. We employ this technique, along with a self-distillation methodology, in the realm of FL in a framework called FjORD.  ...  In large-scale deployments, client heterogeneity is a fact and constitutes a primary problem for fairness, training performance and accuracy.  ...  C.2 Implementation Details FjORD was built on top of PyTorch [54] and Flower [6] , an open-source federated learning framework which we extended to support Ordered, Federated, and Adaptive Dropout and  ... 
arXiv:2102.13451v5 fatcat:q7pty5tgh5hqlpkirxstxjos4y

Personalized Federated Learning through Local Memorization [article]

Othmane Marfoq, Giovanni Neglia, Laetitia Kameni, Richard Vidal
2022 arXiv   pre-print
In order to tackle this limitation, recent personalized federated learning methods train a separate model for each client while still leveraging the knowledge available at other clients.  ...  Federated learning was originally used to train a unique global model to be served to all clients, but this approach might be sub-optimal when clients' local data distributions are heterogeneous.  ...  The authors are grateful to the OPAL infrastructure from Université Côte d'Azur for providing computational resources and technical support.  ... 
arXiv:2111.09360v3 fatcat:wwr5qxd4azcydhhs5es3hmycgu

Better Methods and Theory for Federated Learning: Compression, Client Selection and Heterogeneity [article]

Samuel Horváth
2022 arXiv   pre-print
Federated learning (FL) is an emerging machine learning paradigm involving multiple clients, e.g., mobile phone devices, with an incentive to collaborate in solving a machine learning problem coordinated  ...  In this thesis, we identify several of these challenges and propose new methods and algorithms to address them, with the ultimate goal of enabling practical FL solutions supported with mathematically rigorous  ...  Lane, "FjORD: Fair and accurate federated learning under heterogeneous targets with ordered dropout", in Conference on Neural Information Processing Systems Spotlight (NeurIPS 2021), 2021.  ... 
arXiv:2207.00392v1 fatcat:4ulwzbpusjbntnijrwe4vlgspe

Integrated European Polar Research Programme [article]

Signe Bech ANDERSEN, Torben R CHRISTENSEN, Doris ABELE, Halvor DANNEVIG, Birgit NJÅSTAD, Giovanni MACELLONI
2020 Zenodo  
The European Polar Research Programme is organised in six main chapters dealing with overarching, cross-cutting research issues which should lead to major advances in the understanding of the Polar Regions  ...  current changes and evolutions in the Polar Regions to better definition, prediction and projection of their future conditions, thus supporting adaptation and resilience to the changes.  ...  and machine-learning ensembles.  ... 
doi:10.5281/zenodo.4256023 fatcat:u75kcdy7avg6bfjbbw5l54udma

Fragments in Connection and Algorithmic Rule: Encoding the Urban Image in Motion

Christiane Wagner
2022 Zenodo  
Media, information, and communication technologies offering experiences to individuals in their urban reality construction choices are discussed, faced with the image's paradoxes in its figurative and  ...  This study will also address an understanding in information and communication sciences about virtual realities and effects generated by the world wide web through hypermedia and multimedia (Nelson and  ...  ACKNOWLEDGEMENTS ACKNOWLEDGEMENTS With kind thanks to Dr Dionisio González and Noémie Goudal for allowing copyright permission to reproduce the images within this paper.  ... 
doi:10.5281/zenodo.6503017 fatcat:iwebxttjgbcc5joe7bsmcmbcfe

EDIȚIA a VI-a VIth edition

Sorin Ivan, Dan Postolea, Brîndușa Covaci
2012 Osterreichish Rumanischer Akademischer Verein   unpublished
In the new paradigm of higher education and research, creativity and innovation are fundamental elements.  ...  Through academic education and scientific research, the university produces and transfers knowledge, develops skills in various specializations.  ...  Further research directions: We intend to focus more over the versatile feature of these strategic options; we shall focus on facts that may favour or disadvantage each of these options, and moreover we  ... 
fatcat:kwip3eqndve7vjujzyjelhw3he

"You need to be double cultured to function here": toward an anthropology of Inuit nursing in Greenland and Nunavut

Helle Møller
2011
It also includes the ability to understand, negotiate, and interact, using at least two ways of being in the world, two ways of learning and teaching, and two ways of perceiving the body, health, and disease  ...  These challenges are compounded by Inuit nurses' and students' need to negotiate the languages and cultures of the nursing field, the Southern systems and their Southern colleagues, with the languages  ...  Stevenson Collection, the Royal Ontario Museum, the Thomas Fisher Rare Book Library at the University of Toronto, the Glenbow Museum Archives in Calgary; in Denmark, at the Archives of the Royal Library and  ... 
doi:10.7939/r39p2wf4t fatcat:yhgclav2bvb3jgu2fcepwhga2a