Sep 27, 2024: aeon Module Meeting (Anomaly Detection)

Present: MM SS TB DT

[MM] Collection anomaly detection/outlier detection
- Part of anomaly_detection, subpackage or its own package (outlier_detection?)
- Suggestion:
  - aeon.anomaly_detection for both with subpackages series (for subsequence AD) and collection (for whole-series AD)
  - many methods can actually be used for both cases with different preprocessing
[SS] online/streaming?
- currently, there is no plan to support streaming
[MM] When to remove the experimental tag?
- visible user base
- algorithm per method family
- notebook and documentation ready
[SS] Should algorithms be tagged with metadata or are the tags just used for functional?
- rather use subpackages
- but subpackages are just required if we have enough algorithms implements
- one could use the algorithm_type-tag for the diff. method families
[SS] Do we want to have individual algorithm classes for algorithms that use the PyODAdapter, such as LOF, IsolationForest?
- just if the algorithm is important and deserves visibility

N/A

N/A

[SS] look through survey papers and look if our issues cover all families (good algorithms)
[SS] create issues for PyODAdapter-instances
[MM] create issue for streaming (no current plans to implement)
[MM] create issue for collection AD
[MM] Update AEP with current state and experimental tag discussion

N/A