prDeep: Robust Phase Retrieval with a Flexible Deep Network
Authors: Christopher Metzler, Phillip Schniter, Ashok Veeraraghavan, Richard Baraniuk
ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We test and validate pr Deep in simulation to demonstrate that it is robust to noise and can handle a variety of system models. In Section 4, we apply pr Deep to simulated data and show that it compares favorably to existing algorithms with respect to computation time and robustness to noise. |
| Researcher Affiliation | Academia | 1Department of Electrical and Computer Engineering, Rice University, Houston, TX 2The Ohio State University, Columbus, OH. |
| Pseudocode | No | The paper describes the algorithms and frameworks used but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | Yes | A public implementations of pr Deep is available at https://github.com/ricedsp/pr Deep. |
| Open Datasets | Yes | In particular, we trained with 300 000 overlapping patches drawn from 400 images in the Berkeley Segmentation Dataset (Martin et al., 2001). |
| Dataset Splits | No | The paper mentions 'validation error' in the context of training the Dn CNN denoiser (a component they used), but it does not specify explicit training/validation/test dataset splits for its own experimental setup (e.g., percentages or sample counts for the simulated data). |
| Hardware Specification | Yes | All algorithms were tested using Matlab 2017a on a desktop PC with an Intel 6800K CPU and an Nvidia Pascal Titan X GPU. |
| Software Dependencies | Yes | All algorithms were tested using Matlab 2017a on a desktop PC with an Intel 6800K CPU and an Nvidia Pascal Titan X GPU. pr Deep and Dn CNN-ADMM used a Mat Conv Net (Vedaldi & Lenc, 2015) implementation of Dn CNN. |
| Experiment Setup | Yes | HIO was run for 1000 iterations. WF was run for 2000 iterations. BM3D-pr GAMP was run for 50 iterations. pr Deep was run for 200 iterations four times; once for each of the denoisers networks (trained at standard deviations 60, 40, 20, and 10). pr Deep s parameter λ, which determines the amount of regularization, was set to σw when dealing with Fourier measurements and 0.1 σw when dealing with CDP measurements. |