Extracting Rare Dependence Patterns via Adaptive Sample Reweighting
Authors: Yiqing Li, Yewei Xia, Xiaofei Wang, Zhengming Chen, Liuhua Peng, Mingming Gong, Kun Zhang
ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirical evaluation of synthetic and real-world datasets comprehensively demonstrates the efficacy of our method. Empirically, we conduct extensive experiments on synthetic and real-world data that demonstrate the efficacy of our method. We apply the proposed testing method to both synthetic and real data to evaluate their performance. |
| Researcher Affiliation | Academia | 1Department of Machine Learning, Mohamed Bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE 2Shanghai Key Lab of Intelligent Information Processing, and School of Computer Science, Fudan University, Shanghai, China 3KLAS and School of Mathematics and Statistics, Northeast Normal University, Changchun, China 4College of Mathematics and Computer, Shantou University, Shantou, China 5School of Mathematics and Statistics, The University of Melbourne, Melbourne, VIC, Australia 6Department of Philosophy, Carnegie Mellon University, Pittsburgh, USA. |
| Pseudocode | Yes | Algorithm 1 Reweighted HSIC (RHSIC) ... Algorithm 2 Rare Dependence PC (RD-PC) |
| Open Source Code | Yes | Codes are available at https://github.com/leeedwina430/RKCIT. |
| Open Datasets | Yes | Sachs Dataset. We apply our RHSIC to a flow cytometry dataset (Sachs et al., 2005)... Financial Dataset. We also apply our method to monthly JPY/USD exchange rates (E) and U.S. federal funds rates (F) from 1990 to 2010, sourced from Federal Reserve Economic Data (FRED). |
| Dataset Splits | Yes | We randomly split it into disjoint training (Dtr) and testing (Dte) data. The split ratio is set to 0.5. |
| Hardware Specification | No | The paper does not explicitly describe the hardware used to run its experiments, such as specific GPU or CPU models. |
| Software Dependencies | No | The paper mentions using Python libraries like 'scipy' and 'causal-learn' and refers to other codebases, but it does not specify any version numbers for these software components. |
| Experiment Setup | Yes | The significance level is set to 0.05. The results are obtained after averaging the values in the 100 tests. We set the number of permutations to 2000 to approximate the null distribution. The hyperparameters in our objective functions (9) are set to λ1 = λ2 =1e-3 for RHSIC and λ1 =1e-6, λ2 =1e-1 for RKCIT. And the ϵ for kernel ridge regression is set to 1e-3. |