BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:3D Driven Explainable AI
DTSTART:20230626T160000
DTEND:20230626T180000
DTSTAMP:20260407T132120Z
UID:04547fdfc3ead67f73af5c925944f3c9da534674937883d380c9f066
CATEGORIES:Conferences - Seminars
DESCRIPTION:Yann Bouquet\nEDIC candidacy exam\nExam president: Prof. Pasca
 l Frossard\nThesis advisor: Prof. Mathieu Salzmann\nCo-examiner: Prof. Ami
 r Zamir\n\nAbstract\nExplainable AI (XAI) methods are being explored to ad
 dress the lack of transparency and interpretability in neural networks. Th
 is research proposes using 3D parameters to generate precise post hoc expl
 anations\, improving user control and assessing the robustness of deep lea
 rning models. By incorporating generative models for 3D data and employing
  techniques like signal decompositions\, the goal is to enhance understand
 ing of computer vision models. The study aims to leverage modular infrastr
 ucture and generative models in the 3D domain to identify semantic failure
  modes and improve control over explanations. We aim our contribution to p
 articipate in the development of counterfactual explanations that can go b
 eyond the limited attributes and datasets taken into account in current me
 thods.\n\nBackground papers\nLeclerc Guillaume\, Hadi Salman\, Andrew Ilya
 s\, Sai Vemprala\, Logan Engstrom\, Vibhav Vineet\, Kai Xiao\, et al. “3
 DB: A Framework for Debugging Computer Vision Models.” In _Advances in N
 eural Information Processing Systems_\, edited by S. Koyejo\, S. Mohamed\,
  A. Agarwal\, D. Belgrave\, K. Cho\, and A. Oh\, 35:8498–8511. Curran As
 sociates\, Inc.\, 2022. \nhttps://proceedings.neurips.cc/paper_files/pape
 r/2022/file/3848bc3112429079af85dedb7d369ef4-Paper-Conference.pdf\n\nJeann
 eret Guillaume\, Simon Loic and Frederic Jurie. "Adversarial Counterfactua
 l Visual Explanations." In _2023 IEEE/CVF Conference on Computer Vision an
 d Pattern Recognition (CVPR)\, 2023 \nhttps://arxiv.org/pdf/2303.09962.pd
 f\n\nHui Ka-Hei\, Ruihui Li\, Jingyu Hu\, and Chi-Wing Fu. “Neural Templ
 ate: Topology-Aware Reconstruction and Disentangled Generation of 3D Meshe
 s.” In _2022 IEEE/CVF Conference on Computer Vision and Pattern Recognit
 ion (CVPR)_\, 18551–61. New Orleans\, LA\, USA: IEEE\, 2022. \nhttps://
 openaccess.thecvf.com/content/CVPR2022/papers/Hui_Neural_Template_Topology
 -Aware_Reconstruction_and_Disentangled_Generation_of_3D_Meshes_CVPR_2022_p
 aper.pdf\n 
LOCATION:BC 229 https://plan.epfl.ch/?room==BC%20229
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR