Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1]
Hybrid Federated Learning for Feature & Sample Heterogeneity: Algorithms and Implementation
Authors: Xinwei Zhang, Wotao Yin, Mingyi Hong, Tianyi Chen
TMLR 2024 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we conducted numerical experiments on multiple image classification data sets to validate the performance of the proposed algorithm. |
| Researcher Affiliation | Academia | Xinwei Zhang EMAIL Department of Electrical and Computer Engineering University of Minnesota; Wotao Yin EMAIL Department of Mathematics University of California, Los Angeles; Tianyi Chen EMAIL Department of Electrical and Computer Engineering Rensselaer Polytechnic Institute; Mingyi Hong EMAIL Department of Electrical and Computer Engineering University of Minnesota |
| Pseudocode | Yes | Algorithm 1 Hybrid Federated Matching Algorithm (Hy FEM) ... Algorithm 2 Sub-routines for Algorithm 1 ... Algorithm 3 Model Matching Procedure |
| Open Source Code | No | The paper does not contain an explicit statement about the release of open-source code or a link to a code repository for the described methodology. |
| Open Datasets | Yes | Dataset & data splitting: We consider the Model Net40, Cifar-10, and Euro SAT datasets, the details of which are explained below. We also consider an additional multi-modal dataset, we refer the readers to Appendix C.3 for details. Model Net40 (Wu et al., 2015): Model Net40 is a multiview object classification dataset that has 12 views from different angles as 12 feature blocks for each object. ... Cifar-10 (Krizhevsky, 2009): Cifar-10 is an image classification dataset... Euro SAT (Helber et al., 2019): Euro SAT is a land cover classification satellite image dataset... Heri Graph (Bai et al., 2022) dataset is a multi-modal dataset for heritage site classification. |
| Dataset Splits | Yes | In the training phase of each task, we manually assign a few feature blocks and classes to each client, so that the clients have partially overlapping features and samples and exhibit FH and SH. The settings are summarized in Table 2. It is worth pointing out that in setting Model Net40:2, 12.08% of the data have never been used by any of the clients during training, and in all settings, there is no feature or sample that is shared by all clients, so VFL and HFL algorithms cannot be applied. We conduct two sets of experiments on Model Net40 dataset where setting 1 uses d0 = 4 views and setting 2 uses full d0 = 12 views. ... In the testing phase, the clients evaluate their model on all testing samples with corresponding feature blocks used in the training phase. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper mentions using 'MLP model' and 'CNN part of Res Net-18'/'Res Net-34' for feature extractors, but does not provide specific version numbers for any programming languages, libraries, or frameworks used in the implementation. |
| Experiment Setup | Yes | Training settings: In the experiments, we use the MLP model with one hidden layer as the classifier fm(wm; ). We use the CNN part of Res Net-18 followed by one pooling layer as the feature extractors for the Cifar-10 and Euro SAT datasets; we use the CNN part of Res Net-34 followed by one pooling layer as the feature extractors for Model Net40 dataset. ... In all settings, we fix the total number of updates (i.e., T Q = 4096, with T = 128, Q = 32) for fair comparison and tune the learning rate to achieve the optimal performance for each experiment separately. |