Low-Rank Similarity Metric Learning in High Dimensions

Authors: Wei Liu, Cun Mu, Rongrong Ji, Shiqian Ma, John Smith, Shih-Fu Chang

AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The efficacy of the proposed algorithm is demonstrated through experiments performed on four benchmark datasets with tens of thousands of dimensions.
Researcher Affiliation Collaboration IBM T. J. Watson Research Center Columbia University Xiamen University The Chinese University of Hong Kong {weiliu,jsmith}@us.ibm.com cm3052@columbia.edu sfchang@ee.columbia.edu rrji@xmu.edu.cn sqma@se.cuhk.edu.hk
Pseudocode Yes Algorithm 1 Low-Rank Similarity Metric Learning
Open Source Code No The paper does not provide concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described in this paper.
Open Datasets Yes We carry out the experiments on four benchmark datasets including two document datasets Reuters-28 and TDT2-30 (Cai, He, and Han 2011), and two image datasets UIUCSports (Li and Fei-Fei 2007) and UIUC-Scene (Lazebnik, Schmid, and Ponce 2006).
Dataset Splits Yes On Reuters-28 and TDT2-30, we select 5 C up to 30 C samples for training such that each category covers at least one sample; we pick up the same number of samples for cross-validation; the rest of samples are for testing.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes To run our proposed method LRSML, we fix ϵ = 0.1, ρ = 1, and find that τ = 0.01 makes the linearized ADMM converge within T = 1, 000 iterations on all datasets.