An implementation of the Random Cut Forest data structure for sketching streaming data, with support for anomaly detection, density estimation, imputation, and more.
Currently MLCommons is using ObjectOutputStream to serialize object (code link). But ThresholdedRandomCutForestState doesn't implement Serializable, so it's impossible to serialize via ObjectOutputStream.
If we use same way in MLCommons we need to add one more dependency and we need to maintain two serialization ways. So prefer to implement Serializable to make serialization easier.
Currently MLCommons is using
ObjectOutputStream
to serialize object (code link). ButThresholdedRandomCutForestState
doesn't implementSerializable
, so it's impossible to serialize viaObjectOutputStream
.AD is using
protostuff
to serialize.protostuff
has one bug reported in AD https://github.com/opensearch-project/anomaly-detection/issues/263, we fixed by adding local jar dependency as protostuff team has no clear plan to release new version.If we use same way in MLCommons we need to add one more dependency and we need to maintain two serialization ways. So prefer to implement
Serializable
to make serialization easier.