Robust Multi-Object Matching via Iterative Reweighting of the Graph Connection Laplacian

Authors: Yunpeng Shi, Shaohan Li, Gilad Lerman

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the superior performance of our procedure over state-of-the-art methods using both synthetic and real datasets. ... In 6 we compare performance with previous methods. ... 6 Numerical Experiments: Using synthetic and real data, we compared IRGCL-S&P...
Researcher Affiliation Academia Yunpeng Shi* Shaohan Li Gilad Lerman *Program in Applied and Computational Mathematics, Princeton University School of Mathematics, University of Minnesota yunpengs@princeton.edu, {li000743, lerman}@umn.edu
Pseudocode Yes Algorithm 1 CEMP (reformulated) ... Algorithm 2 Iteratively Reweighted Graph Connection Laplacian (IRGCL)
Open Source Code No The paper mentions using 'codes from https://github.com/zju-3dv/ multiway' for *other* methods (Match Lift and Match ALS), but it does not provide a link or an explicit statement about the public availability of the source code for the methodology proposed in *this* paper.
Open Datasets Yes We compare the performance of the different methods on the Willow database [9], which consists of 5 image datasets. ... [9] M. Cho, K. Alahari, and J. Ponce. Learning graphs to match. In Proceedings of the IEEE Interational Conference on Computer Vision, 2013.
Dataset Splits No The paper describes how synthetic data was generated ('We independently sample nc nodes and for each sampled node we independently corrupt its mc incident edges') and how real data was filtered ('To make those datasets well-posed to PS solvers, we only use the relative permutation Xij between the nodes i and j whose incident edges are not completely corrupted'). However, it does not specify any explicit train/validation/test dataset splits (e.g., percentages, sample counts, or references to predefined splits).
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments, such as GPU models, CPU models, or memory specifications.
Software Dependencies No The paper mentions the use of 'codes from https://github.com/zju-3dv/ multiway' for other methods and 'Alex Net [16]' for feature extraction. However, it does not provide specific version numbers for any software components (e.g., programming languages, libraries, frameworks, or solvers) used in their experiments.
Experiment Setup Yes We use the following parameters for IRGCL-S&P: t0=5, tmax=100, βt=min(2t,40), αt=min(1.2t 1,40), λt=t/(t+1) and F(A)=A. We stop the algorithm whenever P(t+1)=P(t).