Cooperative Graphical Models

Authors: Josip Djolonga, Stefanie Jegelka, Sebastian Tschiatschek, Andreas Krause

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we demonstrate the efficacy of our methods empirically. Our experiments indicate the scalability, efficacy and quality of these schemes. (Section 6: Experiments)
Researcher Affiliation Academia Josip Djolonga Dept. of Computer Science, ETH Z urich josipd@inf.ethz.ch Stefanie Jegelka CSAIL, MIT stefje@mit.edu Sebastian Tschiatschek Dept. of Computer Science, ETH Z urich stschia@inf.ethz.ch Andreas Krause Dept. of Computer Science, ETH Z urich krausea@inf.ethz.ch
Pseudocode Yes Figure 2: Inference with Frank-Wolfe, assuming that VAR-INFERENCE guarantees an upper bound. 1: procedure FW-INFERENCE(f, θ) ... 1: procedure LINEAR-ORACLE(f, τ)
Open Source Code Yes We use the same setting, data and models as [1], as implemented in the pycoop5 package. (footnote 5: https://github.com/shelhamer/coop-cut.)
Open Datasets No For the image segmentation task, the paper states: 'We use the same setting, data and models as [1], as implemented in the pycoop5 package.' However, it does not provide a direct link, DOI, or specific instructions for accessing this dataset within the paper.
Dataset Splits No The paper does not provide specific details on training, validation, or test dataset splits (e.g., percentages or sample counts) for any of its experiments.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions software components such as 'lib DAI [34]', 'cvxpy [35]', 'SCS [36]', and a 'max-flow solver [37]'. However, it does not provide specific version numbers for these software dependencies, which are required for reproducible software description.
Experiment Setup Yes For synthetic experiments: 'The unary potentials were sampled as θi(xi) Uniform( α, α). The edges E were randomly split into five disjoint buckets E1, E2, . . . , E5, and we used f(y) = P5 j=1 hj(y Ej), where y Ei are the coordinates of y corresponding to that group, and the functions {hj} will be defined below.' and 'hi(y Ei) = wi q P |Ei|, with weights wi Uniform(0, β).' and 'θi,j(xi, xj) = wi,j Jxi = xj K, where wi,j Uniform( β, β).'