How belkacem-it/sdn-malware-detection-deeplearning works
This repository presents an SDN malware-detection project with two documented classification approaches: semantic ALBERT and temporal GRU. The tracked model files contain imports and confidentiality notices, not executable pipelines. The README describes flow capture, representations, preprocessing, and real-time detection; those stages are shown as documented capabilities, not verified runtime wiring.
Main entry points: albert_classifier.py ALBERT classifier · gru_classifier.py GRU classifier
Components 10
Flow inputs
- Wireshark capture
- Flow aggregation
- Flow-to-text
- Tokenization
- Temporal features
- Feature normalization
Classifiers
- ALBERT classifier
models/albert_classifier.py - GRU classifier
models/gru_classifier.py
Other components
- Network traffic
- Malware detection
Connections 10
- Network traffic → Wireshark capture: captured by
- Wireshark capture → Flow aggregation: aggregated as flows
- Flow aggregation → Flow-to-text: described as text
- Flow-to-text → Tokenization: tokenized for
- Tokenization → ALBERT classifier: classified by
- Flow aggregation → Temporal features: represented as sequences
- Temporal features → Feature normalization: normalized
- Feature normalization → GRU classifier: classified by
- ALBERT classifier → Malware detection: detects
- GRU classifier → Malware detection: detects