Research group at Budapest University of Technology and Economics, focused on applied machine learning for network traffic analysis and security.
- Encrypted traffic analysis — VPN detection, QUIC classification, malware detection over encrypted channels
- Federated learning — privacy-preserving traffic classification under non-IID and temporally volatile conditions
- Anomaly detection — ML-based network security, dataset integrity, service degradation detection
- Network measurement — flow-level compression, programmable data planes, traffic classification
| Repository | Description |
|---|---|
| MalwareDet-JA4vsFlowStats | ECH-resilient malware detection via flow-level statistical features |
| wavelet-vpn-detection | Binary VPN traffic detection using wavelet features and ML |
| VPN-nonVPN-Dataset | VPN vs. non-VPN traffic dataset |
| Repository | Description |
|---|---|
| IFLforTFC | Incremental federated learning for traffic flow classification |
| FL-QUIC-TC | Federated QUIC traffic classification |
| privacy-preserving-federated-learning | Framework for experimenting with privacy-preserving mechanisms in FL |
| Repository | Description |
|---|---|
| CyberML-DataQuality | Evaluating ML-based anomaly detection across datasets of varied integrity |
| CyberML-CompleteVsFirstN | Early-stage anomaly detection: complete vs. partial flows |
| ServDeg-Dataset | Latency-induced service degradation: methodology and dataset |
| ServDeg-Inter | Inter-flow service degradation detection |
| Repository | Description |
|---|---|
| AutoFlow | Autoencoder-based IP flow record compression |
| P4toNFV | Offloading from P4 switches to NFV in programmable data planes |
| AGFA | Adaptive gradual flow aggregation |
| ml-flow-class-tutorial | Traffic flow classification using ML — tutorial |
Group lead: Adrián Pekár — Google Scholar