Results 91 to 100 of about 166,795 (258)

Local Differential Privacy Is Not Enough: A Sample Reconstruction Attack Against Federated Learning With Local Differential Privacy

IEEE Transactions on Information Forensics and Security
Reconstruction attacks against federated learning (FL) aim to reconstruct users' samples through users' uploaded gradients. Local differential privacy (LDP) is regarded as an effective defense against various attacks, including sample reconstruction in FL, where gradients are clipped and perturbed.
Zhichao You, Xuewen Dong, Shujun Li, Ximeng Liu, Siqi Ma, Yulong Shen   +5 more
openaire   +2 more sources

Cathodic Cage Plasma Deposition of Nanostructured Cu–Fe–Se Coatings on Poly(methyl Methacrylate)

Advanced Engineering Materials, EarlyView.
Nanostructured Cu–Fe–Se coatings are deposited on PMMA by a modified cathodic cage plasma process, enabling low‐temperature deposition on polymer substrates. A transition from discontinuous to compact morphology is observed with temperature, with optimal properties at 200°C, where improved CuFeSe2‐type bonding, lowest sheet resistance, and favorable ...
V. S. S. Sobrinho, M. Naeem, T. H. C. Costa, M. S. Libório, J. C. A. Queiroz, C. B. B. Luna, R. R. M. Sousa, P. S. Vieira, E. M. Araújo   +8 more
wiley   +1 more source

Research on federated learning approach based on local differential privacy

Tongxin xuebao, 2022
As a type of collaborative machine learning framework, federated learning is capable of preserving private data from participants while training the data into useful models.Nevertheless, from a viewpoint of information theory, it is still vulnerable for ...
Haiyan KANG, Yuanrui JI
doaj   +2 more sources

Raman Microscopy at the Organic–Inorganic Interfaces in Human Calcified Aortic Valves Shows the Co‐Existence of Whitlockite Crystals and Carbonated Hydroxyapatite‐Mineralized Collagen Fibrils

Advanced Functional Materials, EarlyView.
This study advances our understanding of aortic valve stenosis by capturing spatially resolved chemical and structural changes at the nanoscale. The findings highlight the potential of combined Raman and electron microscopy for understanding calcification mechanisms across diverse tissue types.
Robin H. M. Van der Meijden, Carolien T. A Kuster, Amber van Broekhoven, Raphaela Allgayer, Luco Rutten, Rona Roverts, Marta Cerruti, Elena Macías‐Sánchez, Jan Hein Cornel, Salloua el Messaoudi, Anat Akiva, Nico Sommerdijk   +11 more
wiley   +1 more source

Learning from Data with Heterogeneous Noise using SGD

, 2014
We consider learning from data of variable quality that may be obtained from different heterogeneous sources. Addressing learning from heterogeneous data in its full generality is a challenging problem.
Chaudhuri, Kamalika, Sarwate, Anand D., Song, Shuang   +2 more
core  

Domain Wall Rebounds Driven by Competing Entropic and Spin‐Transfer Torques in Cylindrical Nanowires

Advanced Functional Materials, EarlyView.
Domain‐wall motion in cylindrical magnetic nanowires driven by nanosecond current pulses. Low current densities efficiently displace domain walls, whereas higher currents cause rebound at the wire ends. The effect results from the interplay between spin‐transfer torque and thermally induced processes, highlighting the role of thermal gradients in ...
Elias Saugar, Jorge Marqués‐Marchán, Sara Catalano, Agustina Asenjo, Michael Foerster, Miguel Angel Niño, Manuel Vazquez, Fèlix Casanova, Arantxa Fraile Rodríguez, Rafael P. Del Real, Oksana Chubykalo‐Fesenko, Cristina Bran   +11 more
wiley   +1 more source

Trajectory-aware privacy-preserving method with local differential privacy in crowdsourcing

EURASIP Journal on Information Security
In spatial crowdsourcing services, the trajectories of the workers are sent to a central server to provide more personalized services. However, for the honest-but-curious servers, it also poses a challenge in terms of potential privacy leakage of the ...
Yingcong Hong, Junyi Li, Yaping Lin, Qiao Hu, Xiehua Li   +4 more
doaj   +1 more source

Speckle‐Engineered Upconversion Amplification in Nanoemulsion‐Templated Hydrogel Microdomes

Advanced Functional Materials, EarlyView.
Nanoemulsion‐confined PEGDA microdomes generate speckle‐like excitation fields that strongly amplify upconversion luminescence upon dehydration, enabling filter‐free visible readout with reversible on–off switching. DMD‐based lithography yields scalable, shape‐programmable arrays for moisture‐responsive displays and optical encryption.
Chaeyeong Ryu, Byungcheon Yoo, Seunghun Lee, Hyungyo Jeong, Sanggyun Jeong, Heesu Cho, Jongwon Oh, Dahye Baek, Jongkyu Kim, Yubin Lee, Eunsu Gu, Dong‐Hwan Kim, Jung‐Hoon Park, Jiseok Lee   +13 more
wiley   +1 more source

Privacy-Preserving Graph Embedding based on Local Differential Privacy

Proceedings of the 33rd ACM International Conference on Information and Knowledge Management
Graph embedding has become a powerful tool for learning latent representations of nodes in a graph. Despite its superior performance in various graph-based machine learning tasks, serious privacy concerns arise when the graph data contains personal or sensitive information.
Zening Li, Rong-Hua Li, Meihao Liao, Fusheng Jin, Guoren Wang   +4 more
openaire   +2 more sources

Metachromatic Butterfly Bile Pigments for Multi‐Level Optical Security Films

Advanced Functional Materials, EarlyView.
Bio‐derived optical security materials are manufactured by embedding butterfly‐based pigments in polymer films. Tunable color and fluorescence responses arise from concentration‐controlled metachromasy, enabling spatially encoded patterns with distinct visible, UV‐active, and spectral signatures.
Limin Wang, Bodo D. Wilts
wiley   +1 more source