A Linear-Time Kernel Goodness-of-Fit Test

Authors: Wittawat Jitkrittum, Wenkai Xu, Zoltan Szabo, Kenji Fukumizu, Arthur Gretton

NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In experiments, the performance of our method exceeds that of the earlier linear-time test, and matches or exceeds the power of a quadratic-time kernel test. In experiments (Section 5), our new linear-time test is able to detect subtle local differences between the density p(x), and the unknown q(x) as observed through samples.
Researcher Affiliation Academia Wittawat Jitkrittum Gatsby Unit, UCL wittawatj@gmail.com Wenkai Xu Gatsby Unit, UCL wenkaix@gatsby.ucl.ac.uk Zoltán Szabó CMAP, École Polytechnique zoltan.szabo@polytechnique.edu Kenji Fukumizu The Institute of Statistical Mathematics fukumizu@ism.ac.jp Arthur Gretton Gatsby Unit, UCL arthur.gretton@gmail.com
Pseudocode No The paper describes algorithms and procedures in prose and mathematical formulas, but it does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/wittawatj/kernel-gof.
Open Datasets Yes We consider crime data from the Chicago Police Department, recording n = 11957 locations (latitude-longitude coordinates) of robbery events in Chicago in 2016.3 ... Data can be found at https://data.cityofchicago.org.
Dataset Splits Yes We divide the sample {xi}n i=1 into two disjoint training and test sets, and use the training set to compute \ FSSD2 \hat\sigma_{H1}+\gamma , where a small regularization parameter \gamma > 0 is added for numerical stability. All tests with optimization use 20% of the sample size n for parameter tuning.
Hardware Specification No The paper does not specify any hardware details such as CPU/GPU models, memory, or specific computing environments used for the experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers (e.g., Python 3.8, PyTorch 1.9, TensorFlow X.Y) that were used in the experiments.
Experiment Setup Yes We evaluate the following six kernel-based nonparametric tests with α = 0.05, all using the Gaussian kernel. All tests with optimization use 20% of the sample size n for parameter tuning. For FSSD tests, we use J = 5.