MPGL: An Efficient Matching Pursuit Method for Generalized LASSO
Authors: Dong Gong, Mingkui Tan, Yanning Zhang, Anton van den Hengel, Qinfeng Shi
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments demonstrate the superior efficiency and accuracy of the proposed method over the state-of-the-arts in both classification and image processing tasks. We conduct the experiments on both synthetic data and real-world data to verify the effectiveness of our algorithm (MPGL). |
| Researcher Affiliation | Academia | Dong Gong, Mingkui Tan, Yanning Zhang, Anton van den Hengel, Qinfeng Shi School of Computer Science and Engineering, Northwestern Polytechnical University, Xi an, China School of Computer Science, The University of Adelaide, Australia, The Australian Centre for Robotic Vision School of Software Engineering, South China University of Technology, China |
| Pseudocode | Yes | Algorithm 1: Cutting Planes for the QCLP Problem (5); Algorithm 2: Finding the Most Violated Constraint; Algorithm 3: ADMM for Solving Subproblem (9); Algorithm 4: Matching Pursuit for Generalized LASSO. |
| Open Source Code | No | The paper states: 'We implement the main scheme of MPGL in Algorithm 4 in Matlab, and that of the Algorithm 3 in C++ with a mex-file interface.' This describes the implementation but does not explicitly state that the code is open-source or provide a link. |
| Open Datasets | Yes | We conduct the experiments on both synthetic data and real-world data to verify the effectiveness of our algorithm (MPGL). ... four real medical datasets for binary classification: Array CGH dataset (Stransky et al. 2006); Leukemia dataset (Golub et al. 1999); Brain Tumor dataset (Nutt et al. 2003); Prostate Cancer dataset (Petricoin et al. 2002). ... A dataset consisting of four examples is studied, in which all images are 256 256 (n = 65,536). |
| Dataset Splits | Yes | We record the leave-one-out accuracy and the runtime for one training process in Table 1. |
| Hardware Specification | Yes | All the experiments are performed on an Intel i5 CPU with 8G RAM. |
| Software Dependencies | No | The paper mentions 'Matlab' and 'C++ with a mex-file interface' for implementation, but no specific version numbers for these or any other software dependencies are provided. |
| Experiment Setup | Yes | In this experiment, we fix the parameter λ in (2) as 0.005 and test all algorithms on the data with different values of n. ... We set λ = 0.001 for GL methods, and use default settings for other methods. ... we set κ as the number of elements in β0 larger than ζ β0 . In practice, ζ 0.5 works well. |