BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Towards Robust and Reliable Model Explanations - Hima Lakkaraju\, 
 Harvard University 
DTSTART:20210504T121500Z
DTEND:20210504T131500Z
UID:TALK155227@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:"Join us on Zoom":https://zoom.us/j/99166955895?pwd=SzI0M3pMVE
 kvNmw3Q0dqNDVRalZvdz09\n\nAs machine learning black boxes are increasingly
  being deployed in domains such as healthcare and criminal justice\, there
  is growing emphasis on building tools and techniques for explaining these
  black boxes in an interpretable manner. Such explanations are being lever
 aged by domain experts to diagnose systematic errors and underlying biases
  of black boxes. In this talk\, I will present some of our recent research
  that sheds light on the vulnerabilities of popular post hoc explanation t
 echniques such as LIME and SHAP\, and also introduce novel methods to addr
 ess some of these vulnerabilities. More specifically\, I will first demons
 trate that these methods are brittle\, unstable\, and are vulnerable to a 
 variety of adversarial attacks. Then\, I will discuss two solutions to add
 ress some of the vulnerabilities of these methods – (i)  a framework bas
 ed on adversarial training that is designed to make post hoc explanations 
 more stable and robust to shifts in the underlying data\; (ii) a Bayesian 
 framework that captures the uncertainty associated with post hoc explanati
 ons and in turn allows us to generate explanations with user specified lev
 els of confidences. I will conclude the talk by discussing results from re
 al world datasets to both demonstrate the vulnerabilities in post hoc expl
 anation techniques as well as the efficacy of our aforementioned solutions
 .\n\n*BIO*: *Hima Lakkaraju* is an Assistant Professor at Harvard Universi
 ty focusing on explainability\, fairness\, and robustness of machine learn
 ing models.  She has also been working with various domain experts in crim
 inal justice and healthcare to understand the real world implications of e
 xplainable and fair ML. Hima has recently been named one of the 35 innovat
 ors under 35 by MIT Tech Review\, and has received best paper awards at SI
 AM International Conference on Data Mining (SDM) and INFORMS. She has give
 n invited workshop talks at ICML\, NeurIPS\, AAAI\, and CVPR\, and her res
 earch has also been covered by various popular media outlets including the
  New York Times\, MIT Tech Review\, TIME\, and Forbes. For more informatio
 n\, please visit: https://himalakkaraju.github.io/
LOCATION:Zoom
END:VEVENT
END:VCALENDAR