Detection and Localization of Changes in Conditional Distributions
Authors: Lizhen Nie, Dan Nicolae
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | This section investigates performance in synthetic data. We report representative results on different forms of X, Y and types of changes, with additional results included in the Appendix. Baselines. We consider three baselines: one existing (the fixed design CP method [31]), and two adapted from existing abrupt CP methods for unpaired data (denoted by DXY and DY ). Localization comparisons are reported in Table 1a. |
| Researcher Affiliation | Academia | Lizhen Nie Department of Statistics The University of Chicago lizhen@uchicago.edu Dan Nicolae Department of Statistics The University of Chicago nicolae@statistics.uchicago.edu |
| Pseudocode | Yes | Algorithm 1 KCE to solve task II (conditional expectation change) Algorithm 2 KCD to solve Task I (conditional distribution change) |
| Open Source Code | Yes | Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] See supplemental material. |
| Open Datasets | Yes | All data are downloaded from https://www.marketwatch.com/. The data we use are collected by [10], where the yields of three-month T-bills are treated as market interest rates. |
| Dataset Splits | No | The paper uses n0 and n1 parameters to define a search range for the change point, but does not explicitly mention train/validation/test dataset splits with percentages or sample counts. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running the experiments. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers (e.g., library names like PyTorch, TensorFlow, or scikit-learn with their specific versions). |
| Experiment Setup | Yes | All bandwidths used for all methods are tuned among Sh = {0.001, 0.01, 0.1, 1, 10} on 10 independently generated data sets. We set FX = N(0, 1), F 0 = N(0, 1), n = 1000, = 0.7, = 700, 0 = 0.05, 1 = 0.95. |