Continuous Latent Process Flows
Authors: Ruizhi Deng, Marcus A. Brubaker, Greg Mori, Andreas Lehrmann
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our model on synthetic data sampled from known stochastic processes to verify its ability to capture a variety of continuous dynamics. We compare our proposed architecture against several baseline models with continuous dynamics that can be used to fit irregular time-series data, including CTFP, latent CTFP, latent SDE, and latent ODE. We also run experiments with Variational RNN (VRNN) [9], a model that can be used to fit sequential data. All implementation and training details can be found in the supplementary material. Our results on synthetic data are displayed in Table 1. |
| Researcher Affiliation | Collaboration | Ruizhi Deng1,2 Marcus A. Brubaker1,3,4 Greg Mori1,2 Andreas M. Lehrmann1 1Borealis AI 2Simon Fraser University 3York University 4Vector Institute |
| Pseudocode | No | The paper does not contain any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | Yes | Code available at https://github.com/Borealis AI/continuous-latent-process-flows |
| Open Datasets | Yes | We evaluate our model on synthetic data sampled from known stochastic processes... Mujoco-Hopper [32], Beijing Air-Quality Dataset (BAQD) [35] and PTB Diagnostic Database (PTBDB) [4, 16] are examples of such datasets. |
| Dataset Splits | No | The paper does not explicitly provide details on train/validation/test splits, only mentioning training/testing data. For example, in Section 5.2: "We create training/testing data at irregular times by drawing time stamps from a Poisson process and mapping them to the nearest observed sample points (see Appendix D for details)." No explicit validation split is specified in the main text. |
| Hardware Specification | No | The paper does not explicitly describe the hardware used for running its experiments. |
| Software Dependencies | No | The paper mentions software components implicitly (e.g., PyTorch via the link to code that would likely use it), but it does not specify any software dependencies with version numbers. |
| Experiment Setup | No | The paper mentions that "All implementation and training details can be found in the supplementary material" but does not provide specific experimental setup details like hyperparameters in the main text. |