My Publications

You can also browse my Google Scholar profile.

  • Vithursan Thangarasa, Mahmoud Salem, Kevin Leong, Shreyas Saxena, Joel Hestness, Sean Lie
    MediSwift : Efficient Sparse Pre-trained Biomedical Language Models
    Under Review, 2024.
    [BibTeX]

      
  • Vithursan Thangarasa*, Shreyas Saxena*, Abhay Gupta, Sean Lie
    Sparse-IFT: Sparse Iso-FLOP Transformations
    Under Review, 2024.
    [BibTeX]

      
  • Vithursan Thangarasa*, Shreyas Saxena*, Abhay Gupta, Sean Lie
    Sparse-IFT: Sparse Iso-FLOP Transformations (earlier version)
    NeurIPS Workshop on Advancing Neural Network Training, 2023.
    [BibTeX] [PDF]

    @inproceedings{thangarasa2023sparse,
      title={Sparse Iso-FLOP Transformations for Maximizing Training Efficiency},
      author={Thangarasa, Vithursan and
              Saxena, Shreyas and 
              Gupta, Abhay and 
              Lie, Sean},
      booktitle={Workshop on Advancing Neural Network Training: Computational Efficiency, Scalability, and Resource Optimization (WANT@NeurIPS 2023)},
      year={2023},
      url={https://openreview.net/forum?id=iP4WcJ4EX0}
      } 
    
  • Vithursan Thangarasa, Abhay Gupta, William Marshall, Tianda Li, Kevin Leong, Dennis DeCoste, Sean Lie, Shreyas Saxena
    SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models
    Conference on Uncertainty in Artificial Intelligence (UAI), 2023.
    [BibTeX] [PDF]

    @inproceedings{thangarasa2023spdf,
      title={SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models},
      author={Thangarasa, Vithursan and 
              Gupta, Abhay and 
              Marshall, William and 
              Li, Tianda and 
              Leong, Kevin and 
              DeCoste, Dennis and 
              Lie, Sean and 
              Saxena, Shreyas},
      booktitle={Conference on Uncertainty in Artificial Intelligence},
      year={2023},
      url={https://openreview.net/forum?id=BywV1Cris4}
      }
    
  • Vithursan Thangarasa, Abhay Gupta, William Marshall, Tianda Li, Kevin Leong, Dennis DeCoste, Sean Lie
    SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models (earlier version)
    ICLR Workshop on Sparsity in Neural Networks, 2023.
    [BibTeX] [PDF] [Poster]

    @inproceedings{thangarasa2023spdfworkshop,
    title={SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models},
    author={Thangarasa, Vithursan and 
            Gupta, Abhay and 
            Marshall, William and 
            Li, Tianda and 
            Leong, Kevin and 
            DeCoste, Dennis and 
            Lie, Sean and 
            Saxena, Shreyas},
    booktitle={ICLR Workshop on Sparsity in Neural Networks},
    year={2023},
    }
    
  • Vitaliy Chiley, Vithursan Thangarasa, Abhay Gupta, Anshul Samar, Joel Hestness, Dennis DeCoste
    RevBiFPN : The Fully Reversible Bidirectional Feature Pyramid Network
    Conference on Machine Learning and Systems (MLSys), 2023.
    [BibTeX] [PDF]

    @inproceedings{mlsysRevBiFPN2023,
      author={Vitaliy Chiley and
              Thangarasa, Vithursan and
              Gupta, Abhay and
              Samar, Anshul and
              Hestness, Joel and
              DeCoste, Dennis},
      title={RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network},
      booktitle={Conference on Machine Learning and Systems (MLSys)},
      year={2023},
      }
    
  • Vitaliy Chiley, Vithursan Thangarasa, Abhay Gupta, Anshul Samar, Joel Hestness, Dennis DeCoste
    RevBiFPN : The Fully Reversible Bidirectional Feature Pyramid Network (earlier version)
    ICML Hardware-Aware Efficient Training (HAET) Workshop, 2022.
    [BibTeX] [PDF] [Project Page]

    @inproceedings{haetRevBiFPN2022,
      author={Chiley, Vitaliy and
              Thangarasa, Vithursan and
              Gupta, Abhay and
              Samar, Anshul and
              Hestness, Joel and
              DeCoste, Dennis},
      title={RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network},
      booktitle={ICML Workshop on Hardware Aware Efficient Training (HAET)},
      year={2022},
      }
    
  • Mihir Pendse, Vithursan Thangarasa, Vitaliy Chiley, Ryan Holmdahl, Joel Hestness, Dennis DeCoste
    Memory Efficient 3D U-Net with Reversible Mobile Inverted Bottlenecks for Brain Tumor Segmentation
    International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) BrainLes, 2021.
    [BibTeX] [PDF]

    @inproceedings{miccaiThangarasa21,
      author={Pendse, Mihir and 
              Thangarasa, Vithursan and
              Chiley, Vitaliy and
              Holmdahl, Ryan and
              Hestness, Joel and
              DeCoste, Dennis},
      title={Memory Efficient 3D U-Net with Reversible Mobile Inverted Bottlenecks for Brain Tumor Segmentation},
      booktitle="Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries",
      year={2021},
      }
    
  • Vithursan Thangarasa, Thomas Miconi, G. W. Taylor
    Enabling Continual Learning with Differentiable Hebbian Plasticity
    International Joint Conference on Neural Networks (IJCNN), 2020.
    [BibTeX] [PDF]

    @inproceedings{ijcnnThangarasa20,
      author={Vithursan Thangarasa and
              Thomas Miconi and
              Graham W. Taylor},
      title={Enabling Continual Learning with Differentiable Hebbian Plasticity},
      booktitle={2020 International Joint Conference on Neural Networks, {IJCNN}},
      year={2020}
    }
    
  • Vithursan Thangarasa, Chuang-Yung Tsai, G. W. Taylor, Urs Köster
    Reversible Fixup Networks for Memory-Efficient Training
    Neural Information Processing Systems (NeurIPS) Systems for ML (MLSys) Workshop, 2019.
    [BibTeX] [PDF] [Project Page]

    @inproceedings{Thangarasa2019RevUp,
      title={Reversible Fixup Networks for Memory-Efficient Training},
      author={Thangarasa, Vithursan and 
            Tsai, Chuang-Yung and 
            Taylor, Graham W and
            Köster, Urs},
      booktitle={Neural Information Processing Systems (NeurIPS): Systems for ML (MLSys) Workshop},
      year={2019},
      url={http://learningsys.org/neurips19/assets/papers/42_CameraReadySubmission_neurips_2019.pdf},
    }
    
  • Vithursan Thangarasa
    Differentiable Hebbian Consolidation for Continual Lifelong Learning
    Master's Thesis, University of Guelph, 2019.
    [BibTeX] [PDF] [Project Page]

      @inproceedings{Thangarasa2019Thesis,
        title={Differentiable Hebbian Consolidation for Continual Lifelong Learning},
        author={Thangarasa, Vithursan},
        year={2019},
        url={https://atrium.lib.uoguelph.ca/xmlui/handle/10214/17440},
      }
      
  • Vithursan Thangarasa, Thomas Miconi, G. W. Taylor
    Differentiable Hebbian Plasticity for Continual Learning (earlier version)
    International Conference on Machine Learning (ICML) Adaptive and Multitask Learning: Algorithms & Systems (AMTL) Workshop, 2019.
    [BibTeX] [PDF] [Project Page]

    @inproceedings{Thangarasa2019DHPforCL,
      title={Differentiable Hebbian Plasticity for Continual Learning},
      author={Thangarasa, Vithursan and 
              Miconi, Thomas and 
              Taylor, Graham W},
      booktitle={International Conference on Machine Learning (ICML):
                   Adaptive and Multitask Learning (AMTL) Workshop},
      year={2019},
      url={https://openreview.net/forum?id=r1x-E5Ss34},
    }
    
  • Vithursan Thangarasa, G. W. Taylor
    Self-Paced Learning with Adaptive Deep Visual Embeddings
    British Machine Vision Conference (BMVC), 2018.
    [BibTeX] [PDF] [Project Page]

    @inproceedings{Thangarasa2018SelfPacedLW,
      title={Self-Paced Learning with Adaptive Deep Visual Embeddings},
      author={Thangarasa, Vithursan and 
              Taylor, Graham W},
      booktitle={British Machine Vision Conference (BMVC)},
      year={2018},
      url={http://bmvc2018.org/contents/papers/1033.pdf},
    }