Machine Learning Attacks - Mitre Atlas Actions

[gdbassett]

To prepare for ML attacks, we may want to start adding actions associated with attacking machine learning models to VERIS. Many of the actions against ML models are normal actions just applied against a different asset, (see issue 439). Thankfully various people have already started to catalog attacks on machine learning including Mitre Atlas (https://atlas.mitre.org/) or Ram's paper: https://arxiv.org/abs/1911.11034.

That said, this is a lot of action varieties to potential add. We may not want to do that yet until we have enough data to actually use the actions enough to show up beyond 'other' or in statistically significant amounts.