Daemons-Q/qsh-satellite-telemanom-ckks-student

CKKS-linear telemetry anomaly student

logistic regression over channel x timestep windows

Part of QSMPC-QKD-QHE-AI-Hybrid, a quantum-safe orchestration demo. This is the encrypted-tier model for the satellite_comms use case: it is a distilled student small enough to evaluate under real CKKS, and the demo runs it that way. It is not the plaintext teacher, and its numbers should be read against that teacher rather than against the published literature.

Measured performance

metric	value
agreement	0.998677
disagreement_on_positives	0.142857
metric_delta	-0.003234
metric_delta_pp	-0.3234
n_eval	307601
n_scoring_units	2
student_metric	0.003234
student_params	49
teacher_metric	0.0
teacher_params	80486

Published baselines this is measured against

• Target metric: event-wise F0.5
• Baseline to beat: 0.968 - ESA-ADB Supplementary Table 9, Mission1 lightweight (ch. 41-46), Telemanom-ESA-Pruned: corrected event-wise F0.5 0.968 (precision 0.999, recall 0.862) for detection of ANOMALIES ALONE - which is what this project scores, so this is the like-for-like row. Earlier rounds cited Table 2's 0.786 instead. That number is real and correctly attributed, but Table 2 scores ALL EVENTS excluding communication gaps, i.e. anomalies PLUS annotated rare nominal events, so it was the wrong comparator and the lenient one. Verified against the paper (arXiv:2406.17826 v2) in Round 24. THREE protocol differences remain and none of them are closed by changing the number: (1) ESA-ADB's metric multiplies event precision by the TIME-LEVEL true-negative rate, which is why most of its baselines read <0.001 - this project records that corrected value as corrected_event_wise_f0_5 beside the uncorrected one; (2) ESA-ADB splits each mission in half, this project tests on the last 30 percent; (3) ESA-ADB's Mission1 lightweight test partition holds 65 annotated events of which 29 are anomalies, against 3 in the window scored here. So this remains an indicative target, and the sample-size gap is roughly tenfold.
• Companion metric shown alongside: event_wise_f0_5_full_channels - reported together because the aggregate figure can look healthy while the class that matters is not.

Training data

• Dataset: ESA Anomaly Dataset (ESA-ADB), Mission1 lightweight channels 41-46
• Licence: CC-BY-3.0-IGO
• Source: https://zenodo.org/records/12528696 (licence read 2026-08-03)
• Attribution (CC-BY-3.0-IGO): Kotowski et al., 'European Space Agency Benchmark for Anomaly Detection in Satellite Telemetry', arXiv:2406.17826. Data: ESA Anomaly Dataset, Zenodo record 12528696, CC BY 3.0 IGO.

Missions are anonymised Mission1/2/3 — they are not named Sentinel anything. Multi-year telemetry; the exact span is NOT stated by the Zenodo record, the paper or the upstream README, so it is withheld here rather than asserted - see DATA_LICENSES.md. 1.19% annotated. Mission3 is excluded upstream for triviality. The CC BY 3.0 IGO term appears on the Zenodo landing page but NOT in its REST metadata, which returns no licence at all - an API-only check reads this corpus as unlicensed.

Notes and limitations

The one published model with no browser evaluation sample, and the reason is the metric rather than the weights. Round 21 gave the other three students samples; this one is scored on event-wise F0.5, which is defined over contiguous runs of a timeline and cannot be computed from a shuffled set of independent rows -- which is exactly what the browser evaluator scores. Shipping a per-row metric instead would verify a number nobody published for it. It needs no scaler, unlike the other three: scripts/train/satellite_telemanom.py fits it on raw flat_te (the last 8 timesteps x 6 channels), so the absent input block in its weights.json is correct rather than the omission it looks like. Worth reading alongside its numbers: the teacher scores 0.0 event-wise F0.5 on this split and the student 0.0032, so agreement of 0.9987 is agreement about finding almost nothing.

Honest scope

This model is published as part of a research proof of concept, not as a production system. Numbers above are what this repository measured on the split described, with the code in scripts/train/. Where a figure is carried from the literature rather than measured here, it is labelled as such.