Multiple Graph Matching and Clustering via Decayed Pairwise Matching Composition

Authors: Tianzhe Wang, Zetian Jiang, Junchi Yan1660-1667

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show the proposed methods achieve excellent trade-off on the traditional multi-graph matching case, and outperform in both matching and clustering accuracy, as well as time efficiency.
Researcher Affiliation Academia Tianzhe Wang, Zetian Jiang, Junchi Yan Shanghai Jiao Tong University {usedtobe, maple jzt, yanjunchi}@sjtu.edu.cn Junchi Yan is the correspondence author and he is also affiliated with Mo E Key Lab of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University.
Pseudocode Yes Algorithm 1: Decayed Pairwise Matching Composition for Multi-cluster Multi-graph Matching
Open Source Code No Source code will be made publicly available.
Open Datasets Yes 2) Real-world dataset. Willow-Object Class (Cho, Alahari, and Ponce 2013) contains images from Caltech-256 and PASCAL VOC2007, which are categorized into 5 classes: 109 Face, 66 Winebottle, 50 Duck, 40 Car and 40 Motorbike.
Dataset Splits No The paper describes generating synthetic datasets and using real-world datasets, but it does not specify explicit training, validation, or testing splits for model development or evaluation, which are common in supervised learning tasks. The evaluation focuses on matching and clustering accuracy on the generated/selected data rather than a validation set for hyperparameter tuning or a test set for generalization.
Hardware Specification Yes Note that all the experiments are performed on a laptop with 2.60GHZ 4-core CPU and 12G memory.
Software Dependencies No The paper mentions using specific algorithms like RRWM (Cho, Lee, and Lee 2010), Match Opt (Yan et al. 2015a), and CAO (Yan et al. 2016a) as baselines and components, but it does not provide version numbers for any software libraries, programming languages, or specific tools used in their implementation.
Experiment Setup Yes Table 1: Parameter setting details for experiments. Inlier #: ni, outlier #: no, deform: υ, k-neighbor: k, sensitivity: σ2, reweight: β, cluster #: nc, cluster size: ng.