A Communication Efficient Stochastic Multi-Block Alternating Direction Method of Multipliers
Authors: Hao Yu
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Figures 1 plots the distance to x versus the computation round index or the communication round index in a log-log scale. It also plots baseline curves 1/t 1 β corresponding to O(1/ϵβ) convergence proven in the theorems. Note that in a log-log scale, curves 1/t 1 β become straight lines with slopes 1 β . That is, if our algorithm has the proven convergence rate, the error curves should be eventually parallel to corresponding baseline for large t. In Figures 1, we observe the numerical result is consistent with our theoretical rate proven in our theorems. This simple experiment verifies the correctness of our theorems. |
| Researcher Affiliation | Industry | Hao Yu Amazon eeyuhao@gmail.com |
| Pseudocode | Yes | Algorithm 1 Two-Layer Communication Efficient ADMM |
| Open Source Code | No | No explicit statement providing access to the source code for the methodology (e.g., a repository link or a statement about code release) was found. |
| Open Datasets | No | The paper describes how the data for experiments was generated or synthetically defined (e.g., 'ci N( ci, σ2 i I)', 'Each feature vector aij is generated from a standard normal distribution', 'generate the label bij = sign(a T ijxtrue + ni)') but does not provide access information (link, citation for public dataset, etc.) to a specific, fixed dataset used for training. In Section 4.1, they use synthetic data based on specified distributions, and in Section 4.2, they generate a problem instance similar to [4] with defined parameters and distributions. |
| Dataset Splits | No | No specific details regarding training, validation, or test dataset splits (e.g., percentages, sample counts, or explicit splitting methodology) were found for the experiments described in the paper. The datasets used in the experiments are synthetically generated or defined by parameters. |
| Hardware Specification | Yes | In an experiment over a machine with a multi-core Intel Xeon Processor E5-2682 2.5GHz. |
| Software Dependencies | Yes | Our multi-core implementation of Algorithm 1 uses Python 3.7 and MPI4PY. |
| Experiment Setup | Yes | For all T 1, if we choose any fixed ρ(t) = ρ > 0, ν(t) = ν 8ρ A 2, K(t) = K T in Algorithm 1 |