Causal Inference for Multi-Fault Satellite Failures
Add acknowledgements section to README
Added acknowledgements section detailing the use of NASA Telemanom framework and datasets for evaluation.
Thank you @khundman for making Telemanom open source, it wouldn't have been possible to benchmark Aethelix's performance against recognized spacecraft anomaly sets :)
+11
+11
README.md
+11
README.md
···
343
343
344
344
---
345
345
346
+
## Acknowledgements
347
+
348
+
- Aethelix uses the **NASA Telemanom** framework as a primary benchmark for evaluating diagnostic accuracy on spacecraft telemetry.
349
+
350
+
- **Datasets:** We evaluate using the SMAP (Soil Moisture Active Passive) and MSL (Mars Science Laboratory) datasets provided by NASA.
351
+
- **Baseline:** Performance is compared against the LSTM-based anomaly detection methods established in the following paper:
352
+
353
+
> Hundman, K., Constantinou, V., Laporte, C., Colwell, I., & Soderstrom, T. (2018). *Detecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholding*. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. https://arxiv.org/abs/1802.04431
354
+
355
+
---
356
+
346
357
## Why Causal Inference?
347
358
348
359
Traditional threshold/correlation-based satellite monitoring fails in multi-fault scenarios: