Observations in the second part of "robust_ai.qmd"

[BravoBaldo]

In "robust_ai.qmd", (Please Note: as always, some parts are personal opinions!)

• "When models are trained on poisoned data, their predictions become reliable and trustworthy." Please review.
• "Case Study 1" at the end of section "17.4.2 Data Poisoning" is the same "Case Study 1" of "14.4.2 Data Poisoning" in "privacy_security"
• "If the model cannot generalize well to these different contexts, its performance may improve." Please review
• Search "[Mahalanobis]" 1) why in square brackets? 2) Why is only "distance" linked?
• Search "[Jensen-Shannon (JS)]" 1) why in square brackets? 2) Why is only "divergence" linked?
• Search "errors in the software-visible state of" the sentence is not clear to me (The dash "-" is misleading!):
  • "Errors in the sw i.e. the visible part of the system state."
  • "Errors in the sw visible to the system state"

Baldo

[profvjreddi]

Thanks @BravoBaldo 🙏

• "Case Study 1" at the end of section "17.4.2 Data Poisoning" is the same "Case Study 1" of "14.4.2 Data Poisoning" in "privacy_security"

This is again because we were migrating some contents away from that chapter into this chapter, and somehow I missed catching this. Thank you!

Fixed the rest!