Variational Bayesian Decision-making for Continuous Utilities

Authors: Tomasz Kuśmierczyk, Joseph Sakaya, Arto Klami

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

Reproducibility Variable Result LLM Response
Research Type Experimental We provide practical strategies for approximating and maximizing the gain, and empirically demonstrate consistent improvement when calibrating approximations for specific utilities. ... We demonstrate the technique in predictive machine learning tasks on the eight schools model [9, 29] and probabilistic matrix factorization on media consumption data.
Researcher Affiliation Academia Tomasz Ku smierczyk Joseph Sakaya Arto Klami Helsinki Institute for Information Technology HIIT Department of Computer Science, University of Helsinki {tomasz.kusmierczyk,joseph.sakaya,arto.klami}@helsinki.fi
Pseudocode No The paper describes algorithms but does not provide structured pseudocode or algorithm blocks.
Open Source Code Yes The code for reproducing all experiments (with additional figures) is available online1. 1https://github.com/tkusmierczyk/lcvi
Open Datasets Yes We demonstrate the technique in predictive machine learning tasks on the eight schools model [9, 29] and probabilistic matrix factorization on media consumption data. ... The eight schools model [9, 29] is a simple Bayesian hierarchical model... We demonstrate LCVI in a prototypical matrix factorization task, modeling the Last.fm data set [3]
Dataset Splits No The paper states: 'We randomly split the matrix entries into even-sized training and evaluation sets', but does not provide specific percentages, sample counts, or citations to predefined splits.
Hardware Specification No The paper mentions 'computational resources' in the acknowledgements but does not provide specific hardware details.
Software Dependencies No The paper mentions software like 'Adam [13]' and 'Stan [5]' but does not provide specific version numbers for these or other software dependencies.
Experiment Setup Yes Whenever not stated differently, we used joint optimization of {h} and λ with Adam [13] (learning rate set to 0.01) ran until convergence (20k epochs for hierarchical model and 3k epochs for matrix factorization with minibatches of 100 rows).