CVPR Daily - Tuesday

Riccardo tells us he has worked previously on domain adaptation and domain generalization , so when he began to explore continual learning, he was keen to work at the intersection between these two topics. “ For me, the fascinating thing about this intersection is to keep adapting a model for its lifespan without redoing unnecessary computation , ” he explains. “ Developing these smart algorithms for which you can avoid forgetting, without the need of storing your whole training set. That’s really fascinating for me! The continual learning community is great, they’re developing so many nice algorithms, and I think they will have a huge impact on the way we deploy machine learning and computer vision models in the real world. ” And how is he feeling about CVPR? “ I really look forward to speaking with the whole CVPR community, ” he smiles. “ It’s just a shame we cannot meet in person, but it’s going to be fun as always! ” To learn more about Riccardo’s work, you can view his pre-recorded presentation and PDF poster at any time, and take part in his live Q&A during Paper Session Four today (Tuesday) at 11:00-13:30 EDT. 12 DAILY CVPR Tuesday Presentation