Muzammal Naseer
Muzammal Naseer

Asst. Professor, Khalifa University

    I received my Ph.D. degree from the Australian National University, Australia in 2020. My thesis was “Novel Concepts and Designs for Adversarial Attacks and Defenses.” I am an assistant professor in the computer science department of the College of Computing and Mathematical Sciences at Khalifa University.

    I am interested in building Robust Intelligent Systems. My research focuses on robust visual-spatial and temporal perception, understanding and explaining AI behavior through adversarial machine learning, representation learning through self-learning ( self-supervision, self-distillation, self-critique, self-reflection), and configuring the role of large language models (LLMs) in building robust AI systems across applications of security and life sciences.

    ⚡ Top-Venue

    • One AAAI 2025 paper accepted
    • One TPAMI 2025 paper accepted.
    • Four CVPR 2024 papers accepted.
    • One ICLR 2024 paper accepted.
    • One AAAI 2024 paper accepted (Oral, Top 9.0%).
    • One NeurIPS 2023 paper accepted.
    • Three ICCV 2023 papers accepted.
    • Three CVPR 2023 papers accepted.
    • One ICLR 2023 paper accepted.
    • One TPAMI 2022 paper accepted.
    • One CVPR 2022 paper accepted (Oral, Top 5.0%).
    • One ICLR 2022 paper accepted (Spotlight, Top 5.0%).
    • One NeurIPS 2021 paper accepted (Spotlight, Top 3.0%).
    • Two ICCV 2021 papers accepted.
    • One CVPR 2020 paper accepted (Oral, Top 5.7%).
    • One NeurIPS 2019 paper accepted.

    🔥 Notable

    • One ACCV 2024 paper accepted (Oral, Top 5.6%, Best Student Paper Runner-up).
    • Four MICCAI 2024 papers accepted.
    • One MICCAI 2023 paper accepted (Early Accept, Top 14.0%).
    • One BMVC 2022 paper accepted (Oral, Top 9.5%).
    • One ACCV 2022 paper accepted (Oral, Top 14.6%).
    Download CV
    📚 My Research

    Use this area to speak to your mission. I’m a research scientist in the Moonshot team at DeepMind. I blog about machine learning, deep learning, and moonshots.

    I apply a range of qualitative and quantitative methods to comprehensively investigate the role of science and technology in the economy.

    Please reach out to collaborate 😃

    Featured Publications
    Geochat: Grounded large vision-language model for remote sensing
    LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts
    S3A: Towards Realistic Zero-Shot Classification via Self Structural Semantic Alignment
    Align Your Prompts: Test-Time Prompting with Distribution Alignment for Zero-Shot Generalization
    Self-regulating Prompts: Foundational Model Adaptation without Forgetting
    See all publications
    Recent Publications
    Kartik Kuckreja, Muhammad Sohail Danish,, Muzammal Naseer, Abhijit Das, Salman Khan, Fahad Shahbaz Khan (2024). Geochat: Grounded large vision-language model for remote sensing. In CVPR.
    Cite Code arXiv
    Hanan Gani, Shariq Farooq Bhat, Muzammal Naseer, Salman Khan, Peter Wonka (2024). LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts. In ICLR.
    Cite Code arXiv
    Sheng Zhang, Muzammal Naseer, Guangyi Chen, Zhiqiang Shen, Salman Khan, Kun Zhang, Fahad Shahbaz Khan (2023). S3A: Towards Realistic Zero-Shot Classification via Self Structural Semantic Alignment. In AAAI.
    Cite Code arXiv
    Jameel Hassan, Hanan Gani, Noor Hussein, Muhammad Uzair Khattak, Muzammal Naseer, Salman Khan, Fahad Shahbaz Khan (2023). Align Your Prompts: Test-Time Prompting with Distribution Alignment for Zero-Shot Generalization. In NeurIPS.
    Cite Code arXiv Project Video Presentation Poster
    Muhammad Uzair Khattak, Syed Talal Wasim, Muzammal Naseer, Salman Khan, Ming-Hsuan Yang, Fahad Shahbaz Khan (2023). Self-regulating Prompts: Foundational Model Adaptation without Forgetting. In ICCV.
    Cite Project Page Code arXiv
    See all publications
    Recent & Upcoming Talks
    Recent News

    Welcome to Wowchemy, the website builder for Hugo

    Welcome 👋 We know that first impressions are important, so we’ve populated your new site with some initial content to help you get familiar with everything in no time.