avatar

Samir Khaki

Machine Learning Researcher / Engineer
University of Toronto
samir.khaki (at) mail.utoronto.ca


About Me

I am a graduate student at the University of Toronto researching efficient machine learning applied to computer vision and natural language processing. Currently at Google building multi-modal machine learning models.

Research Interests

News

[Oct. 2025] Excited to present our paper on efficient VLM inference (SparseVILA) at ICCV’25
[Jun. 2025] Excited to present our paper on efficient fine-tuning (SparseLoRA) at ICML’25
[Jul. 2024] Excited to announce two papers at ECCV 2024 this year!
[Jun. 2024] I was awarded the Vector Scholarship in AI & the Edward S. Rogers Sr. Graduate Scholarships
[Jun. 2024] Delivered an oral presentation for my paper on dataset distillation at CVPR-DD 2024.
[Jan. 2024] My paper on efficient model compression for transformers is accepted to ICLR 2024.
[Dec. 2023] My paper on efficient probabilistic contrastive learning is accepted to ICASSP 2024.
[Jul. 2023] My paper on dataset distillation is accepted to ICCV 2023.

Recent Publications


* Equal Contribution

  1. ICCV
    SparseVILA: Decoupling Visual Sparsity for Efficient VLM Inference
    Samir Khaki, Junxian Guo, Jiaming Tang, Shang Yang, Yukang Chen, Konstantinos N. Plataniotis, Yao Lu, Song Han, Zhijian Liu
    International Conference on Computer Vision (ICCV), 2025.
    PDF Accept (Poster)

  2. ICML
    SparseLoRA: Accelerating LLM Fine-Tuning with Contextual Sparsity
    Samir Khaki*, Xiuyu Li*, Junxian Guo*, Ligeng Zhu, Konstantinos N. Plataniotis, Amir Yazdanbakhsh, Kurt Keutzer, Song Han, Zhijian Liu
    International Conference on Machine Learning (ICML), 2025.
    PDF Code Project Page Accept (Poster)

  3. ECCV
    Data-to-Model Distillation: Data-Efficient Learning Framework
    Ahmad Sajedi*, Samir Khaki*, Lucy Z. Liu, Ehsan Amjadian, Yuri A. Lawryshyn, and Konstantinos N. Plataniotis
    European Conference on Computer Vision (ECCV), 2024.
    PDF Accept (Poster)

  4. ECCV
    Sparse Refinement for Efficient High-Resolution Semantic Segmentation
    Zhijian Liu*, Zhuoyang Zhang*, Samir Khaki, Shang Yang, Haotian Tang, Chenfeng Xu, Kurt Keutzer, and Song Han
    European Conference on Computer Vision (ECCV), 2024.
    PDF Code Project Page Accept (Poster)

  5. ICLR
    The Need for Speed: Pruning Transformers with One Recipe
    Samir Khaki and Konstantinos N. Plataniotis
    International Conference on Learning Representations (ICLR), 2024.
    PDF Code Project Page Accept (Poster)

  6. CVPR-DD
    ATOM: Attention Mixer for Efficient Dataset Distillation
    Samir Khaki*, Ahmad Sajedi*, Kai Wang, Lucy Z. Liu, Yuri A. Lawryshyn, and Konstantinos N. Plataniotis
    IEEE/CVF Conference on Computer Vision and Pattern Recognition: Dataset Distillation Workshop (CVPR-DD), 2024.
    PDF Code Accept (Oral)

  7. ICASSP
    ProbMCL: Simple Probabilistic Contrastive Learning For Multi-label Visual Classification
    Ahmad Sajedi, Samir Khaki, Yuri A. Lawryshyn, and Konstantinos N. Plataniotis
    IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024.
    PDF Code Project Page Accept (Poster)

  8. ICCV
    DataDAM: Efficient Dataset Distillation with Attention Matching
    Ahmad Sajedi*, Samir Khaki*, Ehsan Amjadian, Lucy Z. Liu, Yuri A. Lawryshyn, and Konstantinos N. Plataniotis
    IEEE/CVF International Conference on Computer Vision (ICCV), 2023.
    PDF Code Project Page Accept (Poster)

  9. CVPR-ECV
    CFDP: Common Frequency Domain Pruning
    Samir Khaki, Weihan Luo
    IEEE/CVF Conference on Computer Vision and Pattern Recognition: Efficient Computer Vision Workshop (CVPR-ECV), 2023.
    PDF Code Project Page Accept (Poster)

  10. Arxiv
    End-to-End Supervised Multilabel Contrastive Learning
    Ahmad Sajedi*, Samir Khaki*, Konstantinos N. Plataniotis, and Mahdi S. Hosseini
    Arxiv Pre-Print, 2023.
    PDF Code Early Pre-Print

    11. See full publication list on Google Scholar

Industry Positions

Services

Conference Reviewers


Powered by Jekyll and Minimal Light theme.