Personalized Online Federated Learning with Multiple Kernels

Authors: Pouya M. Ghari, Yanning Shen

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on real datasets showcase the advantages of the proposed algorithm compared with other online federated kernel learning ones.
Researcher Affiliation Academia Pouya M. Ghari University of California Irvine pmollaeb@uci.edu Yanning Shen University of California Irvine yannings@uci.edu
Pseudocode Yes Algorithm 1 The k-th client kernel subset selection at time t. and Algorithm 2 Personalized Online Federated Multi-Kernel Learning (POF-MKL)
Open Source Code Yes Codes are available at https://github.com/pouyamghari/POF-MKL.
Open Datasets Yes real datasets downloaded from UCI machine learning repository [11]: Naval [7], UJI [47], Air [51] and WEC [35].
Dataset Splits No The paper describes data distribution among clients (i.i.d vs non-i.i.d) and total samples (T=500), but does not explicitly provide percentages or counts for training, validation, or test splits.
Hardware Specification Yes all experiments were carried out using Intel(R) Core(TM) i7-10510U CPU @ 1.80 GHz 2.30 GHz processor with a 64-bit Windows operating system.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes Learning rates are set to η = ηk = 1/T, k. Also, exploration rates are set to ξk = 1, k. The dictionary of kernels consists of 51 RBFs with different bandwidth such that the bandwidth of the i-th kernel is σi = 10 2i 52 / 25. The maximum value can be picked for the number of random features D is 100. The maximum number of parameters that a client is allowed to transmit to the server is 1000.