Neural Ordinary Differential Equations for Modeling Epidemic Spreading
Authors: Chrysoula Kosma, Giannis Nikolentzos, George Panagopoulos, Jean-Marc Steyaert, Michalis Vazirgiannis
TMLR 2023 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this paper, we apply the proposed architecture on several network datasets and compare it against state-of-the-art methods under different experimental settings. Our results indicate that the proposed method improves predictions in various spreading scenarios, paving the way for the extensive application of interpretable neural networks in the field of epidemic spreading. Section 3 is explicitly titled "Experiments" and describes the experimental setup, datasets, baselines, and presents quantitative results. |
| Researcher Affiliation | Academia | All authors are affiliated with "École Polytechnique IP Paris, France". The email addresses provided (e.g., EMAIL, EMAIL) use academic domains, indicating an academic affiliation for all authors. |
| Pseudocode | No | The paper describes the proposed GN-ODE model using mathematical equations and textual explanations of its components and operations, but it does not contain any explicitly labeled pseudocode or algorithm blocks in a structured format. |
| Open Source Code | No | The paper does not contain any explicit statements about the release of source code, nor does it provide a link to a code repository or mention code in supplementary materials. |
| Open Datasets | Yes | We perform our experiments on real-world networks that represent social networks and are derived from online social networking and communication platforms (all datasets are publicly available). Specifically, we experiment with the following network datasets: (1) karate (2) dolphins (3) fb-food and (4) fb-social (5) openflights (6) Wiki-Vote (7) Enron (8) Epinions. ... Wiki-Vote, Enron, Epinions are available in https://snap.stanford.edu/data/ and the rest five datasets in https://networkrepository.com/ (Rossi & Ahmed, 2015). |
| Dataset Splits | Yes | The 200 samples were split into training, validation, and test sets with a 60 : 20 : 20 split ratio, respectively. ... Overall, the validation set contains 40 samples, while the test set contains 80 samples. |
| Hardware Specification | Yes | For larger datasets such as Wiki-Vote, Enron, Epinions we only tested the combinations of batch size and hidden dimension size that could fit into the memory of a single GPU (NVidia Quadro RTX 6000). |
| Software Dependencies | No | To solve the system, we utilized the ODE-RK method. ODE-RK follows the implementation of the Sci Py package1 and solves the fixed system of ODEs with a Runge-Kutta solver of order 5(4) (Dormand & Prince, 1980). While SciPy is mentioned, a specific version number for it or other key software dependencies (like Python, PyTorch/TensorFlow) is not provided. |
| Experiment Setup | Yes | We chose learning rate from {0.0001, 0.001, 0.01}, batch size from {2, 4, 8, 16, 32, 64, 128} and hidden dimension size for the trainable layers from {16, 32, 64, 128, 256, 512}. We used the mean absolute error as our loss function and trained each architecture for 500 epochs. For the ODE solver in the case of GN-ODE, we used Euler s method with a step size equal to 0.5. |