Welcome

This is the webpage for the computer vision reading group at Mila - Quebec AI Institute.

In our meetings, we present/discuss papers appearing in the leading computer vision conferences (e.g., CVPR, ICCV, ECCV, …) and machine learning conferences (e.g., NeurIPS, ICML, ICLR, AAAI, …) and also related ideas.

Topics of interest:

Schedule of talks

We will resume our meetings in October. The exact date and time is TBD. If you would like to join or give a talk, send an email!

Date Presenter Title Slides
06/06/2022 Aarash Feizi MaskCOV: A random mask covariance network for ultra-fine-grained visual categorization link
13/06/2022 Oren Kraus Augmenting Convolutional networks with attention-based aggregation link
20/06/2022 Pierre-Luc St-Charles Masked Autoencoders Are Scalable Vision Learners -
27/06/2022 Aarash Feizi Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning link
04/07/2022 Adam Tupper Open-World Semi-Supervised Learning link
11/07/2022 Sabyasachi Sahoo Generalized Out-of-Distribution Detection: A Survey link
18/07/2022 ICML ICML -
25/07/2022 Aarash Feizi Adversarial Masking for Self-Supervised Learning link
08/08/2022 Muawiz Chaudhary ImageNet-trained CNNs are biased towards texture link
15/08/2022 Mats L. Richter Should You Go Deeper? Optimizing Convolutional Neural Network Architectures without Training link
29/08/2022 Andrey Zhmoginov HyperTransformer: Model Generation for Supervised and Semi-Supervised Few-Shot Learning  
12/09/2022 Oscar Mañas MAPL: Parameter-Efficient Adaptation of Unimodal Pre-Trained Models for Vision-Language Few-Shot Prompting  
19/09/2022 Morteza Rezanejad Medial Spectral Coordinates for 3D Shape Analysis  
27/09/2022 Aarash Feizi Efficient Visual Pretraining with Contrastive Detection  
04/10/2022 Samuel Lavoie Simplicial Embeddings in Self-Supervised Learning and Downstream Classification  
11/10/2022 Mats L. Richter Understanding Data Efficient Transformers  
25/10/2022 Randall Balestriero Self-Supervised Learning: From Theory to Best Practices  
01/11/2022 Francesco Paissan Tiny architectures for Tiny Architectures  
08/11/2022 Nicolas Ballas Recent Advances in Low-Shot Classification  
15/11/2022 Olivier J. Hénaff The Virtuous Cycle of Object Discovery and Representation Learning  
22/11/2022 Sebastian Monka Context-driven Visual Object Recognition based on Knowledge Graphs  
13/12/2022 Adrien Bardes VICRegL: Self-Supervised Learning of Local Visual Features  
20/12/2022 Ishan Misra General Purpose Vision Models and Emergent Behaviors  
17/01/2023 Ari Morcos Beyond neural scaling laws: beating power law scaling via data pruning  
31/01/2023 Oren Kraus Self Supervised Models for High Content Microscopy  
07/02/2023 Yuhui Zhang Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learning  
14/02/2023 Yann Dubois Improving Self-Supervised Learning by Characterizing Idealized Representations  
21/02/2023 Ismail Elezi The Unreasonable Effectiveness of Fully-Connected Layers for Low-Data Regimes  
07/03/2023 Aarash Feizi Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations  
14/03/2023 Muawiz Chaudhary Investigating Prediction-Time Batch Normalization Under Label Shift  
28/03/2023 Simon Guiroy Improving Meta-Learning Generalization with Activation-Based Early-Stopping  
11/04/2023 Anas Mahmoud Self-Supervised Image-to-Point Distillation via Semantically Tolerant Contrastive Loss  
18/04/2023 Ankit Goyal RVT: Robotic View Transformer for 3D manipulation  

Contact

This reading group is organized by Aarash Feizi, feel free to contact me at this email!