Multi-Channel Information Field Theory
PUBLIC RESULTS

Public Benchmarks

Reproducible evaluations of MCIFT methods on public datasets. Each card represents a frozen run with its own data split, implementation identity, decision definitions and limitations.

Open the public benchmark repository on GitHub
RUN 01Promising development result

IMS Bearing Early-Warning Benchmark

A derived-score evaluation of an MCIFT early-warning detector on one IMS Set 2 bearing run.

Dataset
IMS Bearing Dataset · IMS Set 2
Domain
Machinery · predictive maintenance
Principal result
4,510 minutes before endpoint, measured from sustained-run start at index 532
Warning confirmation
Index 534
Early calibration crossings
2
Methodology version
ims-set2-v1
Open result
RUN 02Public run report · Under review

NASA IMS Set 2 — six-gate evaluation

Chronological bearing-condition benchmark using separate screening, persistence, progression and conventional-vibration evidence.

Evaluation
536 recordings
Persistent warnings
5
High-confidence warnings
3
Healthy holdout
0/96 full-MCIFT positives
Warning position
49 recordings before terminal record
Localization
Failed

Not necessarily physical-fault-onset lead time.

View Run 02 results

These results are evidence from defined protocols, not scientific or production validation of the complete MCIFT framework.

Contact
Your privacy