On Signal-to-Noise Ratio Issues in Variational Inference for Deep Gaussian Processes

Authors: Tim G. J. Rudner, Oscar Key, Yarin Gal, Tom Rainforth

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Specifically, we show both theoretically and via an extensive empirical evaluation that the SNR of the gradient estimates for the latent variable s variational parameters decreases as the number of importance samples increases.
Researcher Affiliation Academia 1Department of Computer Science, University of Oxford, Oxford, United Kingdon 2Computer Science Department, University College London, London, United Kingdom.
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at https://github.com/ timrudner/snr issues in deep gps.
Open Datasets Yes Dua, D. and Graff, C. UCI machine learning repository, 2017. URL http://archive.ics.uci.edu/ml.
Dataset Splits No The paper states '20 random train-test splits' but does not explicitly mention a separate validation set or its split percentage.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments.
Software Dependencies No The paper mentions open-source code availability but does not specify software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes For these experiments, we consider a two-layer DGP with the hyperparameters that directly affect the SNR the number of importance samples K and the minibatch size set to 50 and 64, respectively.