BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Interpretability in the Wild: When Explainability Meets Causality 
 and Clinical Reality - Prof Sonali Parbhoo (Imperial College London)
DTSTART:20260317T150000Z
DTEND:20260317T160000Z
UID:TALK245767@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:We build models to make better decisions. We build explanation
 s so we can trust those decisions. But what if the explanation is confiden
 tly pointing at the wrong thing? In this lecture I want to show that inter
 pretability is not just a tool for transparency. It is a lens for catching
  when your model has learned something it should not have\, and a bridge t
 o causal reasoning that actually holds up in the real world. I'll start wi
 th a provocation: concept-based models are one of our most powerful tools 
 for interpretability\, but they carry a hidden assumption that is almost a
 lways violated in healthcare data. The representations underlying those co
 ncepts are riddled with shortcuts\, spurious correlations that look predic
 tive during training and collapse at deployment. I'll show why regularizat
 ion\, our default fix\, quietly makes things worse\, and how intervening d
 irectly in gradient space can rescue the concepts we thought we had. Then 
 I'll show what becomes possible when concepts are clean: I'll share how th
 ey can be used to evaluate clinical policies we never actually tested\, le
 tting us ask counterfactual questions about patient care using structure a
  clinician would recognize and trust. If interpretability has ever felt li
 ke a box-ticking exercise to you\, I hope this lecture changes that.
LOCATION:Lecture Theatre 2\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR