Quasi-Monte Carlo Variational Inference

Authors: Alexander Buchholz, Florian Wenzel, Stephan Mandt

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We furthermore provide theoretical guarantees on qmc for Monte Carlo objectives that go beyond mcvi, and support our findings by several experiments on large-scale data sets from various domains.
Researcher Affiliation Collaboration 1ENSAE-CREST, Paris 2TU Kaiserslautern, Germany 3Disney Research, Los Angeles, USA.
Pseudocode Yes Algorithm 1: Quasi-Monte Carlo Variational Inference
Open Source Code No In Appendix D we show how our approach can be easily implemented in your existing code.
Open Datasets Yes As in (Miller et al., 2017), we apply this model to the frisk data set (Gelman et al., 2006) that contains information on the number of stop-and-frisk events within different ethnicity groups.
Dataset Splits No The paper describes the datasets used (e.g., 100-row subsample of wine dataset) but does not provide specific split information like percentages or counts for training, validation, or test sets.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running the experiments.
Software Dependencies No The paper mentions software like 'randtoolbox' (Christophe and Petr, 2015) and 'Adam optimizer' (Kingma and Ba, 2015) but does not provide specific version numbers for these or other software dependencies.
Experiment Setup Yes In the first three experiments we optimize the elbo using the Adam optimizer (Kingma and Ba, 2015) with the initial step size set to 0.1, unless otherwise stated. ... For the score function estimator, we set the initial step size of Adam to 0.01.