Summary of Plot-tal — Prompt Learning with Optimal Transport For Few-shot Temporal Action Localization, by Edward Fish et al.
PLOT-TAL – Prompt Learning with Optimal Transport for Few-Shot Temporal Action Localization
by Edward Fish, Jon Weinbren, Andrew Gilbert
First submitted to arxiv on: 27 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces a novel approach to temporal action localization (TAL) in few-shot learning, addressing limitations of single-prompt methods that often lead to overfitting. The proposed multi-prompt learning framework enhanced with optimal transport learns diverse prompts for each action, capturing general characteristics and distributing the representation to mitigate overfitting risk. Optimal transport theory is employed to efficiently align these prompts with action features, optimizing for a comprehensive representation that adapts to video data’s multifaceted nature. Experiments demonstrate significant improvements in action localization accuracy and robustness on THUMOS-14 and EpicKitchens100 datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper helps us better understand how computers can recognize actions in videos when shown only a few examples. The current methods often fail to adapt to new situations because they rely too heavily on single views or perspectives. To overcome this, the researchers propose a new approach that uses multiple prompts to describe each action, allowing the model to learn more general characteristics and be less likely to overfit. This method is tested on two challenging datasets and shows significant improvements in recognizing actions accurately and robustly. |
Keywords
* Artificial intelligence * Few shot * Overfitting * Prompt
