Frank-Wolfe with Subsampling Oracle
Authors: Thomas Kerdreux, Fabian Pedregosa, Alexandre d’Aspremont
ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We illustrate computational gains on regression problems, involving both 1 and latent group lasso penalties. |
| Researcher Affiliation | Academia | 1D.I., UMR 8548, Ecole Normale Sup erieure, Paris, France. 2UC Berkeley, USA. 3ETH Zurich, Switzerland. 4CNRS, France. |
| Pseudocode | Yes | Algorithm 1: Randomized Frank-Wolfe algorithm, Algorithm 2: Randomized Away-steps FW (RAFW) |
| Open Source Code | No | The paper does not provide any statement or link indicating the availability of open-source code for the described methodology. |
| Open Datasets | Yes | We generate a synthetic dataset following the setting of (Lacoste-Julien & Jaggi, 2015)... On figure 3, we test the Lasso problem on the E2006-tf-idf data set (Kogan et al., 2009)... |
| Dataset Splits | No | The paper does not specify exact split percentages, absolute sample counts, or reference predefined splits with citations for training, validation, or test sets. It mentions synthetic data generation and dataset sizes, but not specific partitioning for reproducibility. |
| Hardware Specification | No | The paper does not describe the specific hardware used to run its experiments, such as CPU/GPU models or memory specifications. It mentions 'wall-clock time' but no hardware details. |
| Software Dependencies | No | The paper does not provide a reproducible description of ancillary software, lacking specific version numbers for libraries, frameworks, or solvers. It does not mention any software details beyond general concepts. |
| Experiment Setup | Yes | We generate a synthetic dataset following the setting of (Lacoste-Julien & Jaggi, 2015), with a Gaussian design matrix A of size (200, 500)... The 1 ball radius set to 40... subsampling parameter η = p |A| = 0.05... d = 10000 we consider a collection G of groups of size 10 with an overlap of 3... The regularizing parameter is β = 14... |