Exact Rate-Distortion in Autoencoders via Echo Noise

Authors: Rob Brekelmans, Daniel Moyer, Aram Galstyan, Greg Ver Steeg

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We proceed to analyse the log-likelihood performance of relevant models on three image datasets: static Binary MNIST [38], Omniglot [24] as adapted by Burda et al. [8], and Fashion MNIST (f MNIST) [44]. All models are trained with 32 latent variables using the same convolutional architecture as in Alemi et al. [3] except with Re LU activations. We trained using Adam optimization for 200 epochs, with an initial learning rate of 0.0003 decaying linearly to 0 over the last 100 epochs. Results are averaged from ten runs of each model after removing the highest and lowest outliers.
Researcher Affiliation Academia Information Sciences Institute University of Southern California Marina del Rey, CA 90292 brekelma, moyerd@usc.edu; galstyan, gregv@isi.edu
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not provide an explicit statement or link to open-source code for the methodology it describes.
Open Datasets Yes We proceed to analyse the log-likelihood performance of relevant models on three image datasets: static Binary MNIST [38], Omniglot [24] as adapted by Burda et al. [8], and Fashion MNIST (f MNIST) [44].
Dataset Splits No The paper mentions training and testing but does not specify explicit training/validation/test dataset splits (e.g., percentages or sample counts) needed for reproduction. While it evaluates on 'test' sets, it does not specify how the data was partitioned for validation.
Hardware Specification No The paper does not explicitly describe the hardware (e.g., specific GPU/CPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions 'Adam optimization' and 'Re LU activations' but does not provide specific version numbers for any software dependencies or libraries.
Experiment Setup Yes All models are trained with 32 latent variables using the same convolutional architecture as in Alemi et al. [3] except with Re LU activations. We trained using Adam optimization for 200 epochs, with an initial learning rate of 0.0003 decaying linearly to 0 over the last 100 epochs.