Non-Euclidean Self-Organizing Maps
Authors: Dorota Celińska-Kopczyńska, Eryk Kopczyński
IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our general experimental setup is as follows... Computing 57600 embeddings takes 4 hours on 8-core Intel(R) Core(TM) i7-9700K CPU @ 3.60GHz. Our implementation is included in the Rogue Viz non-Euclidean geometry engine. Comparison of simulated and Gaussian dispersion. We use the aforementioned measures of quality to check if simulated dispersion improves the quality of the embedding in comparison to Gaussian. |
| Researcher Affiliation | Academia | Dorota Celi nska-Kopczy nska , Eryk Kopczy nski Institute of Informatics, University of Warsaw {dot,erykk}@mimuw.edu.pl |
| Pseudocode | Yes | See Algorithm 1 for the pseudocode which computes Pi,j,t. Algorithm 1 Dispersion algorithm . |
| Open Source Code | Yes | The results of our experiments, code and visualizations are at https://figshare.com/articles/software/Non-Euclidean_Self-Organizing_Maps_code_and_data/16624393. |
| Open Datasets | Yes | To visualize the result of the proposed algorithm we will use the classic iris flower dataset by Fisher [Fisher, 1936] and the palmerpenguins dataset [Horst et al., 2020]. |
| Dataset Splits | No | No explicit training/validation/test dataset splits were specified. |
| Hardware Specification | Yes | Computing 57600 embeddings takes 4 hours on 8-core Intel(R) Core(TM) i7-9700K CPU @ 3.60GHz. |
| Software Dependencies | No | The paper mentions 'Our implementation is included in the Rogue Viz non-Euclidean geometry engine.' but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | We use the following parameters: tmax = 30000 iterations, learning coefficient η = 0.1, dispersion precision p = 10 4, T is the number of dispersion steps until the max value/min value 1.6, 60 landscape dimensions, manifolds with about 520 tiles. |