Flattening the Density Gradient for Eliminating Spatial Centrality to Reduce Hubness

Authors: Kazuo Hara, Ikumi Suzuki, Kei Kobayashi, Kenji Fukumizu, Milos Radovanovic

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Using real-world datasets, we demonstrate that the proposed method improves k-NN classification performance and outperforms an existing hub-reduction method.
Researcher Affiliation Academia National Institute of Genetics Mishima, Shizuoka, Japan; Yamagata University Yonezawa, Yamagata, Japan; The Institute of Statistical Mathematics Tachikawa, Tokyo, Japan; University of Novi Sad Novi Sad, Serbia
Pseudocode No The paper describes the proposed methods mathematically and in prose but does not include structured pseudocode or algorithm blocks.
Open Source Code No The paper mentions using a MATLAB script for mutual proximity, 'We used a MATLAB script norm mp gaussi.m distributed at http://ofai.at/ dominik.schnitzer/mp.', which is an existing method. However, it does not provide concrete access to the source code for the methodology described in this paper developed by the authors.
Open Datasets Yes We used the two datasets from the Kent Ridge Biomedical Dataset Repository, Leukemia and Lung Cancer,7 as well as the two datasets in the UCI machine learning repository, MFeat and ISOLET.8
Dataset Splits Yes We assessed performance according to the accuracy of the prediction using leave-one-out cross-validation.
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 mentions using 'MATLAB script' for a baseline comparison but does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes For the additional parameter κ, we select a value from [1, n 1] such that the hubness is maximally reduced, where n is the dataset size. We assessed performance according to the accuracy of the prediction using leave-one-out cross-validation. (for different k {1, 5, 10, 20})