Michael Kamp
Monash University
43 Papers
122 Citations
Michael Kamp is an academic researcher from Monash University. The author has contributed to research in topics: Computer science & Artificial neural network. The author has an hindex of 9, co-authored 39 publications. Previous affiliations of Michael Kamp include University of Bonn & Fraunhofer Society.
Chat about Author
Papers
•Posted Content
FedBN: Federated Learning on Non-IID Features via Local Batch Normalization.
TL;DR: In this article, the authors propose to use local batch normalization to alleviate the feature shift before averaging models, which achieves faster convergence rate than the classical FedAvg for non-iid data.
254
Efficient Decentralized Deep Learning by Dynamic Model Averaging
Michael Kamp,Linara Adilova,Joachim Sicking,Fabian Hüger,Peter Schlicht,Tim Wirtz,Stefan Wrobel +6 more
- 10 Sep 2018
TL;DR: Kamp et al. as mentioned in this paper proposed an efficient protocol for decentralized training of deep neural networks from distributed data sources, which allows to handle different phases of model training equally well and to quickly adapt to concept drifts.
115
•Posted Content
Efficient Decentralized Deep Learning by Dynamic Model Averaging.
Michael Kamp,Linara Adilova,Joachim Sicking,Fabian Hüger,Peter Schlicht,Tim Wirtz,Stefan Wrobel +6 more
TL;DR: An extensive empirical evaluation validates major improvement of the trade-off between model performance and communication which could be beneficial for numerous decentralized learning applications, such as autonomous driving, or voice recognition and image classification on mobile phones.
102
Communication-efficient distributed online prediction by dynamic model synchronization
Michael Kamp,Mario Boley,Daniel Keren,Assaf Schuster,Izchak Sharfman +4 more
- 15 Sep 2014
TL;DR: The first protocol for distributed online prediction that aims to minimize online prediction loss and network communication at the same time is presented, and it remains applicable when the data is non-stationary and shows rapid concept drift.
Privacy-preserving mobility monitoring using sketches of stationary sensor readings
Michael Kamp,Christine Kopp,Michael Mock,Mario Boley,Michael May +4 more
- 23 Sep 2013
TL;DR: This paper proposes a solution to both tasks using an extension of linear counting sketches to map several individuals to the same position in a sketch, while at the same time the inaccuracies introduced by this overloading are compensated by using several independent sketches.