Large neural models (such as Transformers) achieve state-of-the-art
perf...
Mixup is a regularization technique that artificially produces new sampl...
Negative sampling schemes enable efficient training given a large number...
Deep learning has yielded extraordinary results in vision and natural
la...
We tackle the panoptic segmentation problem with a conditional random fi...
Recent works on zero-shot learning make use of side information such as
...
Pixel-level labelling tasks, such as semantic segmentation, play a centr...
We tackle the problem of optimizing over all possible positive definite
...
We propose a framework for 2D shape analysis using positive definite ker...
Symmetric Positive Definite (SPD) matrices have become popular to encode...
In this paper, we develop an approach to exploiting kernel methods with
...
Modeling videos and image-sets as linear subspaces has proven beneficial...