BScNets: Block Simplicial Complex Neural Networks

Authors: Yuzhou Chen, Yulia R. Gel, H. Vincent Poor6333-6341

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments indicate that BSc Nets outperforms the state-of-the-art models by a significant margin while maintaining low computation costs.
Researcher Affiliation Academia Yuzhou Chen,1 Yulia R. Gel,2 H. Vincent Poor1 1Department of Electrical and Computer Engineering, Princeton University 2Department of Mathematical Sciences, University of Texas at Dallas
Pseudocode No The paper presents an architecture diagram (Figure 2) and describes the model components in text, but it does not include a formal pseudocode block or algorithm.
Open Source Code Yes Our datasets and codes are available on https://github.com/ BSc Nets/BSc Nets.git.
Open Datasets Yes We experiment on three types of networks (i) citation networks: Cora and Pub Med (Sen et al. 2008); (ii) social networks: (1) flight network: Airport (Chami et al. 2019), (2) criminal networks: Meetings and Phone Calls (Cavallaro et al. 2020), and (3) contact networks: High School network and Staff Community (Salath e et al. 2010; Eletreby et al. 2020); (iii) disease propagation tree: Disease (Chami et al. 2019).
Dataset Splits Yes Following (Chami et al. 2019), for all datasets, we randomly split edges into 85%/5%/10% for training, validation, and testing.
Hardware Specification Yes We implement our proposed BSc Nets with Pytorch framework on two NVIDIA RTX 3090 GPUs with 24 Gi B RAM.
Software Dependencies No The paper mentions implementing the model with the 'Pytorch framework' but does not specify its version or any other software dependencies with version numbers.
Experiment Setup Yes For all datasets, BSc Nets is trained by the Adam optimizer with the Cross Entropy Loss function. More details about the experimental setup and hyperparameters are in Appendix B.2.