Product Grassmann Manifold Representation and Its LRR Models
Authors: Boyue Wang, Yongli Hu, Junbin Gao, Yanfeng Sun, Baocai Yin
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our proposed PGLRR and KPGLRR methods on the following public datasets: MNIST Handwritten dataset1, CMU-PIE dataset2, ALOI dataset3, SKIG dataset4, Highway Traffic dataset5. Our methods are assessed against the following clustering methods:... |
| Researcher Affiliation | Academia | Boyue Wang1, Yongli Hu1, Junbin Gao2, Yanfeng Sun1 and Baocai Yin1,3 1Beijing Key Laboratory of Multimedia and Intelligent Software Technology College of Metropolitan Transportation, Beijing University of Technology Beijing, 100124, China boyue.wang@emails.bjut.cn, {huyongli,yfsun,ybc}@bjut.edu.cn 2School of Computing and Mathematics, Charles Sturt University Bathurst, NSW 2795, Australia jbgao@csu.edu.au 3School of Software Technology at Dalian University of Technology, Dalian 116620, China |
| Pseudocode | Yes | Algorithm 1 The whole procedures about Problem (7). Input: The Product Grassmann sample set {[Xi]}N i=1, [Xi] PGn:p1,..,p M and the balancing penalty parameter λ. Output: The Low-Rank Representation Z 1: Initialize:J = Z = 0, A = B = 0, μ = 10 6, μmax = 1010 and ε = 10 8 2: for m=1:M do 3: for i=1:N do 4: for j=1:N do 5: Δm ij tr[(Xm j T Xm i )(Xm T i Xm j )]; 6: end for 7: end for 8: end for 9: for m=1:M do 10: Δ Δ + Δm :: ; 11: end for 12: Performing SVD on Δ Δ UDU T 13: Calculating the coefficient matrix Z by Z UDλU T |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described, nor does it explicitly state that the code is available in a repository or supplementary materials. |
| Open Datasets | Yes | MNIST Handwritten dataset1, CMU-PIE dataset2, ALOI dataset3, SKIG dataset4, Highway Traffic dataset5. |
| Dataset Splits | No | The paper describes constructing 'test datasets' with varying numbers of samples and applying PCA, but does not specify explicit train/validation/test splits with percentages, sample counts, or a detailed cross-validation setup for its own experiments. |
| Hardware Specification | Yes | Our experiments are coded in Matlab 2014a and implemented on a machine with Intel Core i7-4770K 3.5GHz CPU. |
| Software Dependencies | Yes | Our experiments are coded in Matlab 2014a and implemented on a machine with Intel Core i7-4770K 3.5GHz CPU. |
| Experiment Setup | Yes | Our experiments are coded in Matlab 2014a and implemented on a machine with Intel Core i7-4770K 3.5GHz CPU. All color images are converted into gray images and normalized with mean zero and unit variance. ... we apply PCA to reduce the raw vectors to a low dimension which equals to the number of PCA components retaining 95% of its variance energy. ... Algorithm 1 ... Input: The Product Grassmann sample set ... and the balancing penalty parameter λ. ... 1: Initialize:J = Z = 0, A = B = 0, μ = 10 6, μmax = 1010 and ε = 10 8 |