Pitfalls in the use of Parallel Inference for the Dirichlet Process

Authors: Yarin Gal, Zoubin Ghahramani

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present both theoretical and experimental evidence, analysing the load balance for the inference and showing that it is independent of the size of the dataset and the number of nodes available in the parallel implementation. We then ran the inference procedure for 250 iterations initialising the cluster assignments to a single cluster and the parameter α to the true concentration, and evaluated the relative time spent in each node (clock cycles per thread to be exact) using K = 2, 4, 6, 8 nodes.
Researcher Affiliation Academia Yarin Gal YG279@CAM.AC.UK University of Cambridge Zoubin Ghahramani ZOUBIN@ENG.CAM.AC.UK University of Cambridge
Pseudocode No No pseudocode or algorithm blocks were found in the paper.
Open Source Code No The paper refers to implementations by other authors (Chang & Fisher III, 2013) but does not state that the code for this specific paper's methodology is open-source or provide access links.
Open Datasets No The paper states: 'In this experiment we sampled a mega-set (1M points) from a DP mixture of Gaussians with a Gaussian-Wishart prior for parameter values α = 0.5 and α = 2'. However, it does not provide concrete access information (e.g., link, DOI, or a formal citation with authors and year for a public dataset) for these sampled datasets.
Dataset Splits No The paper discusses 'sampled datasets' and 'inference procedure' but does not specify explicit training, validation, or test dataset splits for its experiments.
Hardware Specification No The paper mentions 'on a 12 cores machine' but does not provide specific hardware details such as exact CPU/GPU models, memory, or other detailed computer specifications.
Software Dependencies No The paper mentions using an implementation 'as implemented in (Chang & Fisher III, 2013) with minor bug corrections' but does not specify any software names with version numbers for dependencies.
Experiment Setup Yes We then ran the inference procedure for 250 iterations initialising the cluster assignments to a single cluster and the parameter α to the true concentration, and evaluated the relative time spent in each node (clock cycles per thread to be exact) using K = 2, 4, 6, 8 nodes.