MoTE: Reconciling Generalization with Specialization for Visual-Language to Video Knowledge Transfer
This work addresses a key challenge in video recognition for AI researchers, offering a method to improve both generalization and specialization without significant trade-offs, though it is incremental in nature.
The paper tackles the trade-off between zero-shot generalization and close-set performance in transferring visual-language knowledge to video recognition by introducing MoTE, a framework that balances these aspects through a mixture of temporal experts and regularization techniques, achieving state-of-the-art or competitive results on datasets like Kinetics-400, Kinetics-600, UCF, and HMDB.
Transferring visual-language knowledge from large-scale foundation models for video recognition has proved to be effective. To bridge the domain gap, additional parametric modules are added to capture the temporal information. However, zero-shot generalization diminishes with the increase in the number of specialized parameters, making existing works a trade-off between zero-shot and close-set performance. In this paper, we present MoTE, a novel framework that enables generalization and specialization to be balanced in one unified model. Our approach tunes a mixture of temporal experts to learn multiple task views with various degrees of data fitting. To maximally preserve the knowledge of each expert, we propose \emph{Weight Merging Regularization}, which regularizes the merging process of experts in weight space. Additionally with temporal feature modulation to regularize the contribution of temporal feature during test. We achieve a sound balance between zero-shot and close-set video recognition tasks and obtain state-of-the-art or competitive results on various datasets, including Kinetics-400 \& 600, UCF, and HMDB. Code is available at \url{https://github.com/ZMHH-H/MoTE}.