Minimizing Trust Leaks for Robust Sybil Detection
Authors: János Höner, Shinichi Nakajima, Alexander Bauer, Klaus-Robert Müller, Nico Görnitz
ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our empirical evaluation shows significant advantages of TSR over stateof-the-art competitors on a variety of attacking scenarios on artificially generated data and realworld datasets. and 6. Empirical Evaluation on Synthetic Data and 7. Empirical Evaluation on Real-world Data |
| Researcher Affiliation | Collaboration | J anos H oner 1 2 Shinichi Nakajima 2 3 Alexander Bauer 2 3 Klaus-Robert M uller 2 3 4 5 Nico G ornitz 2 1Math Plan, 10587 Berlin, Germany 2Machine Learning Group, Berlin Institute of Technology, 10587 Berlin, Germany 3Berlin Big Data Center 4Max Planck Society 5Korea University. and JH was supported by Math Plan Gmb H and inno Campus, TU-Berlin. |
| Pseudocode | No | The paper contains mathematical derivations and theoretical explanations, but no explicitly labeled 'Pseudocode' or 'Algorithm' blocks. |
| Open Source Code | No | The paper does not provide any statement about releasing source code, nor does it include a link to a code repository for the described methodology. |
| Open Datasets | Yes | We also evaluated our method on a sample of the Facebook graph Leskovec & Mcauley collected from survey participants using the Facebook app. and Leskovec, Jure and Mcauley, Julian J. Learning to discover social circles in ego networks. In Advances in Neural Information Processing Systems 25, pp. 539 547, 2012. |
| Dataset Splits | No | The paper describes selecting 'labeled examples' or 'trusted nodes' for training/seeding its method but does not provide specific details on how the overall datasets (synthetic or Facebook) were formally split into training, validation, and test sets for model development and evaluation (e.g., percentages or counts for each split). |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used to run the experiments. |
| Software Dependencies | No | The paper mentions methods like the 'Louvian clustering method' and 'BFGS' optimizer, but it does not provide specific version numbers for any software, libraries, or frameworks used in the implementation or experiments. |
| Experiment Setup | Yes | We generated a sample network (|VH| = 200 and |VS| = 30) and select 15 honest nodes and 8 Sybil nodes randomly, which will be used as labeled examples for our TSR. and The trusted nodes |VT | = 50 were randomly distributed among all clusters and a small subset of Sybils |VD| = 30 was chosen as known Sybil nodes. and We examine the following: Wilcoxon-Mann-Whitney (WMW) loss (Yan et al., 2003). Smooth hinge-loss variant A smooth variant of the classical support vector machine hinge-loss with two additional parameters: a decision boundary b 2 R and a scaling parameter a 2 R: and fw( u,v) = (1 exp( w> u,v)) 1. and F(w) = λ 2 kwk2 + P h(p(i)(w), yi). |