Parameter-free Locally Accelerated Conditional Gradients
Authors: Alejandro Carderera, Jelena Diakonikolas, Cheuk Yin Lin, Sebastian Pokutta
ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our theoretical results are complemented by numerical experiments, which demonstrate local acceleration and showcase the practical improvements of PF-La CG over non-accelerated algorithms, both in terms of iteration count and wall-clock time. |
| Researcher Affiliation | Academia | 1Georgia Institute of Technology, Atlanta, GA, USA 2Department of Computer Sciences, University of Wisconsin-Madison, Madison, WI, USA 3Zuse Institute Berlin and Technische Universit at Berlin, Berlin, Germany. |
| Pseudocode | Yes | Algorithm 1 Away-Step Frank-Wolfe Algorithm: AFW(x0, S0)... Algorithm 2 AGD-Iter(yk 1, vk 1, zk 1, Ak 1, ηk, σ, ϵ0, η0)... Algorithm 3 ACC(x0, η0, σ)... Algorithm 4 PF-La CG(x0 vert(X), ϵ > 0) |
| Open Source Code | Yes | Our code can be found at https://github.com/ ericlincc/Parameter-free-La CG. |
| Open Datasets | No | The paper describes generating synthetic datasets for its experiments (e.g., 'M Rn n and b Rn have entries sampled uniformly at random between 0 and 1'), but it does not provide access to a publicly available or open dataset. |
| Dataset Splits | No | The paper does not explicitly provide details about training, validation, or test dataset splits. It describes the characteristics of the generated data but not how it was partitioned for experimentation. |
| Hardware Specification | No | The paper states 'we run each process using one CPU core', but it does not specify the model or type of CPU (e.g., Intel Core i7, AMD Ryzen) or other hardware components used. |
| Software Dependencies | No | The paper mentions implementation in 'Python 3' but does not specify the exact minor version of Python or provide version numbers for any other software libraries or dependencies used. |
| Experiment Setup | Yes | The function being minimized in this example is f(x) = x T M T M + α1n x/2 + b T x, where M Rn n and b Rn have entries sampled uniformly at random between 0 and 1 and n = 10000. The parameter α = 500 is set so that the objective function satisfies m 500. |