Discovering conflicting groups in signed networks
Authors: Ruo-Chun Tzeng, Bruno Ordozgoiti, Aristides Gionis
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our experimental evaluation shows that, compared to state-of-the-art baselines, our methods find solutions of higher quality, are faster, and recover ground-truth conflicting groups with higher accuracy. |
| Researcher Affiliation | Academia | Ruo-Chun Tzeng KTH Royal Institute of Technology rctzeng@kth.se Bruno Ordozgoiti Aalto University bruno.ordozgoiti@aalto.fi Aristides Gionis KTH Royal Institute of Technology argioni@kth.se |
| Pseudocode | Yes | Algorithm 1: SCG (A, k) Spectral Conflicting Group detection; Algorithm 2: Solve-Max-DRQ (A, q) Find maximum discrete Rayleigh quotient; Algorithm 3: Min Angle Round (v, q) Deterministic rounding by minimum-angle heuristic; Algorithm 4: Random Round (v, q) Randomized rounding |
| Open Source Code | Yes | All methods have been implemented in Python 3.1https://github.com/rutzeng/SCG-Neur IPS2020. |
| Open Datasets | Yes | The datasets we have used are all publicly available and the detailed information can be found in Supplementary D.1. (Section 7) |
| Dataset Splits | No | The paper discusses evaluating on synthetic and real-world graphs but does not specify any training, validation, or test dataset splits in terms of percentages, sample counts, or explicit splitting methodologies typically used for model training and evaluation. |
| Hardware Specification | Yes | All the experiments are executed on a machine with Intel Core i5 at 1.8 GHz with 8 GB RAM. |
| Software Dependencies | No | The paper states 'All methods have been implemented in Python 3.' but does not provide specific version numbers for any other software libraries or dependencies used (e.g., NumPy, SciPy, etc.). |
| Experiment Setup | No | The paper details hyperparameters for a baseline method (KOCG) as 'α = 1/(k 1), β = 50, and ℓ= 5000', but it does not provide specific hyperparameter values or detailed training configurations for its own proposed methods (SCG-MA, SCG-MO, SCG-B, SCG-R). |