Federated Online Prediction from Experts with Differential Privacy: Separations and Regret Speed-ups
Authors: Fengyu Gao, Ruiquan Huang, Jing Yang
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6 Numerical Experiments Experiments demonstrate the effectiveness of our proposed algorithms. We conduct experiments in a synthetic environment to validate the theoretical performances of Fed-DP-OPE-Stoch, and compare them with its single-player counterpart Limited Updates (Asi et al., 2022b). We conduct experiments in a synthetic environment, comparing Fed-SVT with the single-player model Sparse-Vector (Asi et al., 2023). |
| Researcher Affiliation | Academia | Fengyu Gao, Ruiquan Huang, Jing Yang School of EECS, The Pennsylvania State University, USA {fengyugao, rzh5514, yangjing}@psu.edu |
| Pseudocode | Yes | Algorithm 1 Fed-DP-OPE-Stoch: Client i, Algorithm 2 Fed-DP-OPE-Stoch: Central server, Algorithm 3 DP-FW at client i (Asi et al., 2021b), Algorithm 4 Fed-DP-OPE-Stoch (CDP): Client i, Algorithm 5 Fed-DP-OPE-Stoch (CDP): Central server, Algorithm 6 Batch algorithm M for 1-Select (Jain et al. (2023), Algorithm 2 with k = 1), Algorithm 7 Fed-SVT: Client i, Algorithm 8 Fed-SVT: Central server |
| Open Source Code | No | The paper does not contain a direct statement or link for open-source code for its proposed methods within the provided text. The NeurIPS Paper Checklist states 'Yes' for providing open access to data and code via supplemental material, but this material is not included in the provided PDF. |
| Open Datasets | Yes | Additionally, we evaluate the performances of Fed-SVT on the Movie Lens-1M dataset (Harper and Konstan, 2015) in Appendix G. |
| Dataset Splits | No | We conduct experiments in a synthetic environment to validate the theoretical performances of Fed-DP-OPE-Stoch, and compare them with its single-player counterpart Limited Updates (Asi et al., 2022b). We conduct experiments in a synthetic environment, comparing Fed-SVT with the single-player model Sparse-Vector (Asi et al., 2023). We use the Movie Lens-1M dataset (Harper and Konstan, 2015) to evaluate the performances of Fed SVT, comparing it with the single-player model Sparse-Vector (Asi et al., 2023). Explanation: The paper mentions using synthetic data and the Movie Lens-1M dataset but does not provide specific train/validation/test split percentages or sample counts. |
| Hardware Specification | Yes | The simulations were conducted on a system with a 2.3 GHz Dual-Core Intel Core i5, Intel Iris Plus Graphics 640 with 1536 MB, and 16 GB of 2133 MHz LPDDR3 RAM. |
| Software Dependencies | No | The paper describes algorithms and experiments but does not provide specific software dependencies with version numbers. |
| Experiment Setup | Yes | We set m = 10, T = 2^14, ε = 10, δ = 0 and d = 100. We set m = 10, T = 2^9, ε = 10, δ = 0 and d = 100. In Fed SVT, we experiment with communication intervals N = 1, 30, 50 |