COMET Flows: Towards Generative Modeling of Multivariate Extremes and Tail Dependence
Authors: Andrew McDonald, Pang-Ning Tan, Lifeng Luo
IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on both synthetic and real-world datasets demonstrate the effectiveness of COMET Flows in capturing both heavytailed marginals and asymmetric tail dependence compared to other state-of-the-art baseline architectures. |
| Researcher Affiliation | Academia | 1Department of Computer Science and Engineering, Michigan State University 2Department of Geography, Environment, and Spatial Sciences, Michigan State University {mcdon499, ptan, lluo}@msu.edu |
| Pseudocode | No | The paper describes the architecture and mathematical formulations of COMET Flows, but it does not include any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | Yes | All code is available at https://github.com/ andrewmcdonald27/COMETFlows. |
| Open Datasets | Yes | CLIMDEX1 extreme climate indices [Sillmann et al., 2013]. Motivated by the projected increase in extreme temperatures and rainfall under climate change, we extract data from a model run under the RCP 8.5 scenario of continued high greenhouse gas emissions, and aim to model the joint distribution of such extremes over North America. We consider the region encompassing 25 N 75 N and 170 W 50 W and simulation data from the years 2091-2100... We use data from 2091-2098 for training, 2099 for validation, and 2100 for testing. A subset of this data was presented in Figure 2. Benchmark Data To contextualize the performance of COMET Flows, we consider the POWER and GAS benchmark datasets originally presented in [Papamakarios et al., 2017]. |
| Dataset Splits | Yes | We generate 200,000 vectors in x R8 for training, 25,000 vectors for validation, and 25,000 for a heldout test set. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory, or specific cloud instances) used for running the experiments. |
| Software Dependencies | No | The paper mentions software components like 'scipy.stats.genpareto.fit' but does not provide specific version numbers for these or any other key software dependencies required for replication. |
| Experiment Setup | No | The paper mentions an early stopping criterion ('trained until validation loss fails to improve for two consecutive epochs') and some details about noise perturbation, but it does not provide specific hyperparameter values such as learning rates, batch sizes, optimizers, or detailed network architectures for the experiments. |