A Bayesian Nonparametrics View into Deep Representations

Authors: Michał Jamroż, Marcin Kurdziel, Mateusz Opala

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental setup. First, we employ DP-GMMs to investigate representational complexity in CNNs that can exploit patterns in data and networks that are forced to memorize random labels. We also compare models with different depths, widths and regularization techniques. To this end, we train several CNN architectures on CIFAR-10 and Mini-Image Net datasets4.
Researcher Affiliation Academia Michał Jamro z AGH University of Science and Technology Krakow, Poland mijamroz@agh.edu.pl Marcin Kurdziel AGH University of Science and Technology Krakow, Poland kurdziel@agh.edu.pl Mateusz Opala AGH University of Science and Technology Krakow, Poland mo@matthewopala.com
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes To this end, we train several CNN architectures on CIFAR-10 and Mini-Image Net datasets4.
Dataset Splits No The paper mentions using a 'test part of the dataset' but does not explicitly provide specific training, validation, or test dataset splits needed for reproduction (e.g., percentages, sample counts, or detailed splitting methodology).
Hardware Specification No The paper mentions 'PL-Grid Infrastructure' in the acknowledgements, but does not provide specific details on the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers).
Experiment Setup No The paper states that details on network architectures, training protocols, and hyperparameters are provided in Appendix C and D, but these specific experimental setup details are not explicitly present in the main text.