Imtiaz Masud Ziko



About


Currently, I am an Associate Principal AI Scientist @ AstraZeneca working towards efficient and effective discovery and delivery of potential medicines using AI, as a part of Center for AI within Data Sciences & AI, BioPharmaceuticals R&D. Previously, I was R&T Lead (AI) @ Thales , where I worked in different projects involving computer vision and machine learning.

I completed my PhD in machine learning from École de Technologie Supérieure (ETS) –University of Quebec, Montréal, Canada. Before that, I completed my Erasmus Mundus Masters in 2014 visiting three universities in Europe: Norwegian University of Science and Technology (NTNU) in Norway, University of Granada in Spain and Jean Monnet University in France.

My research work and interest connects dots around ML/AI for drug discovery, LLMs, Diffusion models, representation learning, few-shot learning, graph based models, out-of-distribution detection for robust ML deployment, continual learning, fairness in machine learning and optimization methods. The motivations of these works are mainly to make AI powered systems to learn and deliver effectively and efficiently with a few or no active supervision while being scalable and robust. I serve as a regular reviewer for the major AI conferences such as the AAAI, ICML, NeurIPS, CVPR etc.


News

  • 2023-11-01: Joined @ AstraZeneca as an Associate Principal AI scientist.
  • 2023-10-24: Our paper on Domain Incremental Learning is accepted at WACV 2024.
  • 2023-07-17: Our paper Generalized Category Discovery is accepted at ICCV 2023.
  • 2021-03-02: Our paper on Few-shot segmentation is accepted at CVPR 2021.
  • 2021-01-04: Started working as a Lead R&T (AI) @ Thales.
  • 2020-12-08: Our paper on fair clustering is accepted at AAAI ‘21.
  • 2020-09-27: Paper accepted at Neurips ‘20.
  • 2020-07-30: Successfully defended my PhD dissertation.
  • 2020-07-02: ECCV paper on metric learning got accepted as spotlight!. Check here.
  • 2020-06-01: ICML paper on few-shot learning got accepted.
  • 2018-09-05: NeurIPS paper got accepted as a spotlight.

Publications

  • MoP-CLIP: A Mixture of Prompt-Tuned CLIP Models for Domain Incremental Learning
    Julien Nicolas, Florent Chiaroni, Imtiaz Ziko, Ola Ahmad, Christian Desrosiers and Jose Dolz
    In WACV, 2023. [PDF]

  • Parametric Information Maximization for Generalized Category Discovery
    Florent Chiaroni, Jose Dolz, Ziko Imtiaz Masud, Amar Mitiche and Ismail Ben Ayed
    In ICCV, 2023. [PDF] [Code]

  • Task Adaptive Feature Transformation for One-Shot Learning
    Imtiaz Masud Ziko, Freddy Lecue and Ismail Ben Ayed. In arXiv, 2023. [PDF]

  • Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?
    Malik Boudiaf, Hoel Kervadec, Imtiaz Masud Ziko, Pablo Piantanida, Ismail Ben Ayed and Jose Dolz
    In CVPR, 2021. [PDF] [Code]

  • Variational Fair Clustering
    Imtiaz Masud Ziko, Jing Yuan, Eric Granger and Ismail Ben Ayed
    In AAAI, 2021. [PDF] [Code]

  • Transductive Information Maximization For Few-Shot Learning
    Malik Boudiaf, Imtiaz Masud Ziko, Jérôme Rony, Jose Dolz, Pablo Piantanida and Ismail Ben Ayed
    In NeurIPS, 2020. [PDF] [Code]

  • Laplacian Regularized Few-Shot Learning
    Imtiaz Masud Ziko, Jose Dolz, Eric Granger, and Ismail Ben Ayed
    In ICML, 2020. [PDF] [Code]

  • Metric learning: cross-entropy vs. pairwise losses
    Malik Boudiaf, Jérôme Rony, Imtiaz Masud Ziko, Eric Granger, Marco Pedersoli, Pablo Piantanida and Ismail Ben Ayed
    In *ECCV    , 2020. (Spotlight) [PDF] [Code]

  • Scalable Laplacian K-modes
    Imtiaz Masud Ziko, Eric Granger and Ismail Ben Ayed
    In NeurIPS, Montreal, Canada, December 2018. (Spotlight) [PDF] [Code]

  • Supervised spectral subspace clustering for visual dictionary creation in the context of image classification
    Imtiaz Masud Ziko, Elisa Fromont, Damien Muselet and Marc Sebban
    In ACPR, Kuala Lumpur, Malaysia, 3-6 Nov. 2015. [PDF]

  • Design and Creation of a Multi-Illuminant Scene Image Dataset
    Imtiaz Masud Ziko, Shida Beigpour and Jon Y. Herdeberg
    In ICISP, Springer, 531-538, July 2014. [PDF] [Dataset]