foundation models / video / vision&language / self-supervised learning / deep learning dynamics / semantic segmentation / object detection / implicit regularization / image / instance segmentation / video object segmentation / temporal action localization / promptable vision models / optimizer / neural tangent kernel / mean-field theory / feature learning / contrastive learning / statistical mechanics / semi-supervised learning / object-centric learning / noisy labels / language / generative models / expressivity / weakly supervised temporal action localization / tracking / test-time training / survey / spatio-temporal action localization / source-free domain adaptation / scaling laws / random matrix theory / probability theory / loss landscape / in-context learning / image captioning / effective field theory / edge of stability / double descent / domain adaptation / diffusion models / depth separation / dataset / convergence theory / class imbalance / active learning / action recognition / Gaussian process