Tabular Classification
Transformers
ONNX
quantum-safe
privacy-preserving
telecom-ids
qsh-telecom-qml-ids / feature_spec.json
Daemons-Q's picture
Quantum-kernel SVC + variational QNN: artefacts + measured metrics
2070c02 verified
Raw
History Blame Contribute Delete
701 Bytes
{
"features": [
"qber_mean",
"qber_std",
"sift_ratio",
"multiphoton_rate",
"timing_jitter",
"dark_count_rate",
"click_asymmetry"
],
"classes": [
"normal",
"intercept_resend",
"photon_number_splitting",
"trojan_horse",
"detector_blinding",
"wavelength_trojan",
"rng_manipulation",
"combined"
],
"note": "8 classes = 7 attack types + normal traffic (ATTACKS[0] == 'normal'), not 'eight attack scenarios'. An earlier revision listed the CLASS LABELS under 'features'; the real model inputs are the QKD telemetry statistics in 'features', from FEATURES in src/qsh/qml_ids/attacks.py. Generated from the source, not written by hand."
}