Conceptualizing Curse of Dimensionality with Parallel Coordinates

Authors: G. Devi, Charu Chauhan, Sutanu Chakraborti

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

Reproducibility Variable Result LLM Response
Research Type Theoretical We report on a novel use of parallel coordinates as a pedagogical tool for illustrating the non-intuitive properties of high dimensional spaces with special emphasis on the phenomenon of Curse of Dimensionality. Also, we have collated what we believe to be a representative sample of diverse approaches that exist in literature to conceptualize the Curse of Dimensionality. We envisage that the paper will have pedagogical value in structuring the way Curse of Dimensionality is presented in classrooms and associated lab sessions.In this paper, we have attempted to collate what we believe to be a representative sample of diverse perspectives that exist in literature for illustrating COD. These perspectives are founded on either mathematical conclusions or statistical comparisons.
Researcher Affiliation Academia G Devi , Charu Chauhan and Sutanu Chakraborti Department of Computer Science and Engineering Indian Institute of Technology Madras, Chennai-600036, Tamil Nadu, India {devsjee, charu.chauhan.8888}@gmail.com, sutanuc@cse.iitm.ac.in
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code No There is no statement or link indicating the release of open-source code for the methodology.
Open Datasets No The paper is conceptual and pedagogical, and does not use datasets for training or evaluation in the traditional sense.
Dataset Splits No The paper is conceptual and pedagogical; it does not involve data splits for validation.
Hardware Specification No The paper does not mention any specific hardware used for its conceptual work or illustrations.
Software Dependencies Yes We have used version 2.2 of XDAT, a free Parallel coordinates software, for our illustrations.
Experiment Setup No The paper is conceptual and does not describe a traditional experimental setup with hyperparameters or training configurations.