Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Continuous Simplicial Neural Networks
Authors: Aref Einizade, Dorina Thanou, Fragkiskos D. Malliaros, Jhony H. Giraldo
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our experiments on real-world datasets demonstrate that COSIMO achieves competitive performance compared to state-of-the-art SNNs in complex and noisy environments. |
| Researcher Affiliation | Academia | Aref Einizade LTCI, Télécom Paris Institut Polytechnique de Paris EMAIL Dorina Thanou EPFL, Lausanne, Switzerland EMAIL Fragkiskos D. Malliaros Centrale Supélec, Inria Université Paris-Saclay EMAIL Jhony H. Giraldo LTCI, Télécom Paris Institut Polytechnique de Paris EMAIL |
| Pseudocode | No | The paper describes methods using mathematical equations and textual explanations, but does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The implementation codes are available in https://github.com/Aref Einizade2/COSIMO. |
| Open Datasets | Yes | To evaluate the effectiveness of COSIMO, we assess its performance on two datasets: a synthetic simplicial complex and the ocean-drifts dataset from [26, 46]. Regression on partial deformable shapes. The Shrec-16 benchmark [47] extends prior mesh classification datasets to meshes with missing parts... Node classification (NC). ...on two benchmark datasets: high-school [48, 49], and senate-bills [48, 50, 51]. Graph classification (GC). ...conducted on the standard proteins dataset [53]. |
| Dataset Splits | Yes | The dataset is divided into a training set (199 shapes) and a test set (400 shapes). We partition each dataset chronologically, reserving the initial 80% for training the encoder and the final 20% for evaluation, as in the literature [44, 52]. Model performance is assessed by computing the average classification accuracy over ten stratified folds [44, 52, 45]. |
| Hardware Specification | Yes | The experiments were conducted on an A100 NVIDIA GPU with 40 GB of memory. |
| Software Dependencies | No | In certain cases, we mostly use Topo Model X [57, 58, 24] to implement previous SOTA methods. For accessing and processing real-world datasets, we employ Torch Topo Net X [58]. While software components are mentioned, specific version numbers for these or other key libraries (like Python or PyTorch) are not provided. |
| Experiment Setup | Yes | Detailed hyperparameter configurations for both synthetic and real-world datasets are provided in the code in https://github.com/Aref Einizade2/COSIMO. We use cross-validation for tuning the possible hyperparameters with the selected values provided in Table 5. Table 5: Hyperparameter details for each dataset; lr and nepochs are the learning rate and number of epochs, respectively. |