Repository used to store and track the SSN (short for Siamese Neural Network) tool. This tool main purpose is to easily experiemnt and train Siamese Networks with different configurations and different features.
This tool has been used to obtain the results presented in the following papers:
- Optimizing Predictive Analytics in 5G Networks Through Zero-Trust Operator-Customer Cooperation presented at NFV-SDN 2023 and awarded as best paper.
- ATELIER: service tailored and limited-trust network analytics using cooperative learning published on OJCOMS,
- Faro: a scalable and reliable outage detection algorithm for IoT Mobile Virtual Network Aggregators, the link will be updaed once the paper will be published and available.
The Repository is separated into two main sections:
- DataGeneration, this folder contains all the tools and information used to produce the datasets used furter for the training of the Neural Network model through the SNN tool (sed in the ATELIER work).
- SNN This folder contains the actual python code of the SNN tool, plus all the configuration objects and post-processing scripts used for both the pubblications previously mentioned
Please notice that due to the size of the datasets used in this projects, such files are not available directly in the repository. But both the dataset generated with the DataGeneration tool (Dkr5G) and the one generated as outcome from the all the training cycles are available on Zenodo
Please refer to the 'reproducibility' section in the wiki for details on how to download and use the mentioned datasets and reproduce the results presented in ATELIER.
Refer to the documentation available on github pages for the SNN tool and examples on how to use it.