A Fast and Provable Algorithm for Sparse Phase Retrieval

Authors: Jian-Feng CAI, Yu Long, Ruixue WEN, Jiaxi Ying

ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments show that our algorithm achieves a significantly faster convergence rate than state-of-the-art methods. ... In this section, we present a series of numerical experiments designed to validate the efficiency and accuracy of our proposed algorithm.
Researcher Affiliation Academia Jian-Feng Cai1,2, Yu Long3 , Ruixue Wen1, Jiaxi Ying1,2 1 Hong Kong University of Science and Technology 2 HKUST Shenzhen-Hong Kong Collaborative Innovation Research Institute 3 Guangxi University
Pseudocode Yes Algorithm 1 Proposed algorithm
Open Source Code Yes Our codes are available at https://github.com/jxying/Sparse PR.
Open Datasets No The paper generates synthetic data for experiments, stating: "The true signal x has s nonzero entries, where the support is selected uniformly from all subsets of [n] with cardinality s, and their values are independently generated from the standard Gaussian distribution N(0, 1)." It does not use or provide access to any publicly available dataset.
Dataset Splits No The paper does not explicitly provide details about training/validation/test splits. It describes generating synthetic data and averaging results over multiple independent trial runs (e.g., "average of 100 independent trial runs") but does not specify data partitioning into train/validation/test sets.
Hardware Specification Yes All experiments were conducted on a 2 GHz Intel Core i5 processor with 16 GB of RAM, and all compared methods were implemented using MATLAB.
Software Dependencies No The paper states that "all compared methods were implemented using MATLAB." However, it does not specify a version number for MATLAB or any other software libraries or dependencies used, which is required for reproducibility.
Experiment Setup Yes We fine-tune the parameters and set: α = 0.7 for Th WF; γ = 0.5, µ = 1 and |I| = m/6 for SPARTA; η = 0.95 for both HTP and our algorithm. The maximum number of iterations for each algorithm is 1000.