Autoencoders that don't overfit towards the Identity

Authors: Harald Steck

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments on three well-known data-sets corroborate the various theoretical insights derived in this paper.
Researcher Affiliation Industry Harald Steck Netflix Los Gatos, California hsteck@netflix.com
Pseudocode No The paper provides mathematical derivations and outlines solution methods like ADMM, but it does not contain explicit pseudocode blocks or algorithms.
Open Source Code Yes The code accompanying these experiments is publicly available at https://github.com/hasteck/EDLAE_Neur IPS2020.
Open Datasets Yes For reproducibility, we follow the experimental set-up in [21], using their publicly available code as well as the same three well-known data-sets Movie Lens 20 Million (ML-20M) [15], Netflix Prize (Netflix) [5], and the Million Song Data (MSD) [6].
Dataset Splits No The paper states it follows the experimental set-up from [21] but does not explicitly provide the train/validation/test split percentages or sample counts within its own text. It mentions evaluating on a "test-set" but does not specify the full data partitioning for training, validation, and testing.
Hardware Specification No The paper mentions that training was "implemented in TensorFlow" and used "Python and Numpy," but it does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) used for running the experiments.
Software Dependencies No The paper mentions "Tensorflow [33]", "Python", and "Numpy". However, it does not provide specific version numbers for any of these software components.
Experiment Setup Yes We determined the optimal training-hyper-parameters (i.e., dropout-probability p and L2-regularization parameter λ) using grid-search.