Solving Random Systems of Quadratic Equations via Truncated Generalized Gradient Flow

Authors: Gang Wang, Georgios Giannakis

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical results demonstrate that: i) The novel orthogonalitypromoting initialization method returns more accurate and robust estimates relative to its spectral counterparts; and ii) even with the same initialization, our refinement/truncation outperforms Wirtinger-based alternatives, all corroborating the superior performance of TGGF over state-of-the-art algorithms.
Researcher Affiliation Academia ECE Dept. and Digital Tech. Center, Univ. of Minnesota, Mpls, MN 55455, USA School of Automation, Beijing Institute of Technology, Beijing 100081, China
Pseudocode Yes Algorithm 1 Truncated generalized gradient flow (TGGF) solvers
Open Source Code Yes The Matlab implementations of TGGF are available at http://www.tc.umn.edu/ gangwang/TAF.
Open Datasets No The paper describes using 'simulated tests under both noiseless and noisy Gaussian models' with i.i.d. ai N(0, In) or ai CN(0, In), which means the data is synthetically generated based on these distributions rather than being a publicly available dataset with a specific link or formal citation.
Dataset Splits No The paper conducts 'Simulated tests under both noiseless and noisy Gaussian models' and states 'Simulated estimates will be averaged over 100 independent Monte Carlo (MC) realizations.' It does not mention traditional train/validation/test splits of a static dataset.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions 'The Matlab implementations of TGGF are available at...', indicating the software used is Matlab, but it does not specify a version number for Matlab or any other software dependencies.
Experiment Setup Yes Algorithm 1 with default values set for pertinent algorithmic parameters. ... take constant step size µ = 0.6/1 for real/complex Gaussian models, truncation thresholds |I0| = 1/6m (the ceil operation), and γ = 0.7. The initial estimate was found based on 50 power iterations, and was subsequently refined by T = 103 gradient-like iterations in each scheme.