Approximate Profile Maximum Likelihood

Authors: Dmitri S. Pavlichin, Jiantao Jiao, Tsachy Weissman

JMLR 2019 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We extensively experiment with the approximate solution, and the empirical performance of our approach is competitive and sometimes significantly better than state-of-the-art performances for various estimation problems. Keywords: Profile maximum likelihood, dynamic programming, sufficient statistic, partition of multi-partite numbers, integer partition
Researcher Affiliation Academia Dmitri S. Pavlichin EMAIL Department of Applied Physics Stanford University Stanford, CA 94305, USA Jiantao Jiao EMAIL Department of Electrical Engineering and Computer Sciences University of California Berkeley, CA 94720, USA Tsachy Weissman EMAIL Department of Electrical Engineering Stanford University Stanford, CA 94305, USA
Pseudocode Yes Algorithm 1 Approximation to level set partition of the approximate D-dimensional PML distribution
Open Source Code Yes We provide code at https://github.com/dmitrip/PML and (Pavlichin et al., 2017) for computing the approximate PML distributions.
Open Datasets No The paper uses synthetic data generated from various distributions (Uniform, Mix 2 Uniforms, Zipf, Geometric) for its experiments. It does not explicitly mention the use of any publicly available or open datasets with specific access information (links, DOIs, formal citations).
Dataset Splits No The paper does not provide explicit details about dataset splits (e.g., train/test/validation percentages or counts). The experiments involve drawing 'n' samples from a distribution for estimation tasks, which does not typically involve predefined dataset splits in the conventional machine learning sense.
Hardware Specification No The paper does not provide any specific hardware details such as GPU models, CPU types, or cloud computing platforms used for running the experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers used for implementing or running the experiments.
Experiment Setup Yes In all cases K = |X| = 104. Uniform is uniform on X, Mix 2 Uniforms is a mixture of two uniform distributions, with half the probability mass on the first K/5 symbols, and the other half on the remaining symbols, and Zipf(α) 1/iα with i {1, . . . , K}. MLE denotes the ML plugin (naive) approach of using the sorted empirical distribution in (61). VV is (Valiant and Valiant, 2013). Each data point represents 100 random trials, with 2 standard error bars smaller than the plot marker for most points.