Conference Presentation

Conference Presentation

Reading group paper presentation

  • Robustness in deep learning:
    • Adversarial robustness
    • Explicit Inductive Bias for Transfer Learning with Convolutional Networks, ICML 2018
    • Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distribution, ICML 2022
    • Robust fine-tuning of zero-shot models, CVPR2022
    • Finetune like you pretrain: Improved finetuning of zero-shot vision models, CVPR2023
    • Spurious Features Everywhere – Large-Scale Detection of Harmful Spurious Features in ImageNet, ICCV2023
    • Mitigating Spurious Correlations in Multi-modal Models during Fine-tuning, ICML 2023
    • Zero-Shot Robustification of Zero-Shot Models, ICLR2024
    • Contrastive Adapters for Foundation Model Group Robustness, NeurIPS2022
    • Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations, ICML2023
  • Recent Meta papers:
    • Scalable Diffusion Models with Transformers, ICCV23
    • Deconstructing Denoising Diffusion Models for Self-Supervised Learning.
    • A Decade’s Battle on Dataset Bias: Are We There Yet?
  • Mixture of Experts
    • Outrageously large neural networks: The sparsely-gated mixture-of-experts layer, ICLR2017
    • Scaling Vision with Sparse Mixture of Experts, NeurIPS2021
    • Learning to Route by Task for Efficient Inference, EMNLP2021
    • Mixture-of-Experts with Expert Choice Routing, NeurIPS2022
    • Multimodal Contrastive Learning with LIMoE: the Language Image Mixture of Experts, NeurIPS2022
    • Towards Understanding the Mixture-of-Experts Layer in Deep Learning, NeurIPS2022
    • Sparse Mixture-of-Experts are Domain Generalizable Learners, ICLR2023
    • M³ViT: Mixture-of-Experts Vision Transformer for Efficient Multi-task Learning with Model-Accelerator Co-design, NeurIPS2022
    • Task-customized Masked Autoencoder via Mixture of Cluster-conditional Experts, ICLR2023
    • The Power of External Memory in Increasing Predictive Model Capacity
    • Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints, ICLR2023
  • Frustratingly Easy Transferability Estimation
    • Frustratingly Easy Transferability Estimation, ICML2022

Talks

Thesis

Online Continual Learning In Image Classification

Committee: Prof. Elias Khalil, Prof. Scott Sanner