Tensor Variable Elimination for Plated Factor Graphs

Authors: Fritz Obermeyer, Eli Bingham, Martin Jankowiak, Neeraj Pradhan, Justin Chiu, Alexander Rush, Noah Goodman

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

Reproducibility Variable Result LLM Response
Research Type Experimental We experiment with plated factor graphs as a modeling language for three tasks: polyphonic music prediction, animal movement modeling and latent sentiment analysis. We report our results in Table 1.
Researcher Affiliation Collaboration 1Uber AI Labs 2Harvard University 3Stanford University.
Pseudocode Yes Algorithm 1 TENSORVARIABLEELIMINATION
Open Source Code Yes Open-source implementations are available; see http://docs.pyro.ai/en/dev/ops.html
Open Datasets Yes Dataset Model JSB Piano Nottingham
Dataset Splits No The paper refers to datasets like JSB Piano, Nottingham, and Sentihood, but does not specify how these datasets were split into training, validation, and test sets (e.g., percentages or sample counts).
Hardware Specification Yes Figure 5 shows results obtained on an Nvidia Quadro P6000 GPU.
Software Dependencies No The paper mentions 'Pyro probabilistic programming language' and 'Num Py', but does not provide specific version numbers for these or other software dependencies.
Experiment Setup No The paper describes model variants and overall experimental design (e.g., '12 different latent variable models'), but it does not provide specific hyperparameter values (e.g., learning rate, batch size, epochs) or detailed training configurations in the main text. It defers some details to supplementary materials.