BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20261115T131237EST-4199MzM07t@132.216.98.100
DTSTAMP:20261115T181237Z
DESCRIPTION:Ivana Malenica \, PhD\n\nAssistant Professor -Department of Bio
 statistics| University of North Carolina\n\nWHEN: Wednesday\, November 18\
 , 2026\, from 3:30 to 4:30 p.m.\n	WHERE: Hybrid | 2001 McGill College Avenu
 e\, Room 1140\; Zoom\n	NOTE: Ivana Malenica will be presenting virtually.\n
 \nAbstract\n\nPlug-in estimators based on flexible machine learning method
 s can exhibit substantial  bias. Classical debiasing methods rely on the e
 fficient influence function (EIF)\, which must be derived and evaluated fo
 r each target parameter\, creating analytical and computational challenges
 . We propose a unified framework for kernel debiased plug-in estimation (K
 DPE)\, which performs debiasing within a reproducing kernel Hilbert space 
 (RKHS) and simultaneously corrects plug-in bias for a broad class of pathw
 ise differentiable parameters. The KDPE constructs the debiased estimator 
 through iterative regularized likelihood updates\, yielding a computationa
 lly tractable procedure that avoids explicit influence-function calculatio
 ns while achieving semiparametric efficiency. We then introduce a refineme
 nt based on the universal least favorable submodel\, termed ULFS–KDPE. Thi
 s method defines a data-adaptive debiasing flow in the RKHS\, formulated a
 s a nonlinear ordinary differential equation on probability densities. We 
 establish existence\, uniqueness\, stability\, and finite-time convergence
  of the empirical score along the induced flow\, and show that the resulti
 ng estimator is regular\, asymptotically linear\, and simultaneously effic
 ient for a broad class of target parameters. Together\, these results prov
 ide a unified\, influence-function-free framework for efficient estimation
  in nonparametric models\, combining theoretical guarantees with computati
 onal tractability. Simulation studies illustrate the methods and support t
 he theoretical findings.\n\nSpeaker Bio\n\nIvana Malenica is an Assistant 
 Professor in the Department of Biostatistics at the University of North Ca
 rolina at Chapel Hill (UNC). Previously\, Dr. Malenica was a Wojcicki-Trop
 er Data Science Fellow at the Harvard Data Science Initiative and a Postdo
 ctoral Fellow in the Department of Statistics at Harvard University. She e
 arned her Ph.D. in Biostatistics from the University of California\, Berke
 ley. Before that\, she studied mathematics and was a fellow at the Transla
 tional Genomics Research Institute. Her methodological research lies at th
 e intersection of causal inference\, non- and semi-parametric statistics\,
  high-dimensional inference\, efficiency theory\, and machine learning. Sh
 e is particularly interested in challenges arising in longitudinal and str
 uctured dependent settings\, including adaptive sequential experiments\, o
 nline learning\, reinforcement learning\, and broader applications in pers
 onalized health.\n\nhttps://sph.unc.edu/adv_profile/ivana-malenica/\n
DTSTART:20261118T203000Z
DTEND:20261118T213000Z
SUMMARY:Kernel Debiased Plug-in Estimation: Simultaneous Debiasing Without 
 Influence Functions
URL:https://www.mcgill.ca/epi-biostat-occh/channels/event/kernel-debiased-p
 lug-estimation-simultaneous-debiasing-without-influence-functions-375636
END:VEVENT
END:VCALENDAR