A Highly Scalable Parallel Algorithm for Isotropic Total Variation Models

Authors: Jie Wang, Qingyang Li, Sen Yang, Wei Fan, Peter Wonka, Jieping Ye

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate that our approach outperforms existing state-of-the-art algorithms, especially in dealing with high-resolution, megasize images.
Researcher Affiliation Collaboration 1Arizona State University, Tempe, AZ 85287 USA 2Huawei Noahs Ark Lab, Hong Kong, China 3King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
Pseudocode Yes Algorithm 1 FAD (Fast ADMM for TV Models)
Open Source Code No The paper does not provide an explicit statement or link for the open-source code of their proposed FAD/pFAD method.
Open Datasets No The paper mentions using synthetic images and specific real images (e.g., moon image, MRI images) but does not provide concrete access information (link, DOI, repository, formal citation with authors/year for a public dataset) for these datasets.
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) for training, validation, or test sets.
Hardware Specification Yes Thus, we implement p FAD with Open MP for this experiment and test all of the algorithms on a server with four quad-core (16 processors in total) Intel Xeon 2.93GHz CPUs and 65GB memory.
Software Dependencies No The paper mentions software like Open MP, MPI, MFISTA, Split Bregman, FCSA, and Matlab, but does not provide specific version numbers for these software components or other dependencies.
Experiment Setup Yes We keep γ fixed (γ = 10) in this paper. For schemes about varying γ, we refer readers to (Boyd et al., 2011). ... We set ϵ = 10 4. ... The regularization parameter λ is set to be 0.2. ... The regularization parameter λ of p FAD is set to be 0.001. ... We use the default settings of the FCSA package. The sample ratio is about 20% and Gaussian noise N(0, 0.012) is added to the undersamples to generate Y. ... We run both FCSA-p FAD and FCSA-MFISTA for 50 iterations.