Research map / Video / Action recognition
Probabilistic Representations for Video Contrastive Learning
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Represent clips as Gaussian distributions and combine them into a video-level mixture | Self-supervised video representation learning · UCF101 and HMDB51 video retrieval | R@1: 67.6 vs 66.9 on UCF101; 40.1 vs 36.4 on HMDB51 | Helped |
| Add probabilistic-distance positive and negative pair mining (ablation) | ProViCo · UCF101 action recognition | accuracy: 84.2 -> 86.1 | Helped |
| Add probabilistic-distance positive and negative pair mining (ablation) | ProViCo · UCF101 video retrieval | R@1: 61.2 -> 64.7 | Helped |
| Increase the number of sampled embeddings from 5 to 7 to 10 (ablation) | ProViCo · UCF101 action recognition | accuracy: 81.2 -> 83.1 -> 83.7 | Helped |
| Increase β from 10^-4 to 10^-3 or 10^-2 (ablation) | ProViCo · UCF101 action recognition | accuracy: 83.7 -> 83.4; 83.7 -> 81.6 | Hurt |
| Set β to 10^-5 (ablation) | ProViCo · UCF101 action recognition | accuracy | Hurt |
| Increase the number of clips used to form the video distribution from 1 to 4 (ablation) | ProViCo · UCF101 action recognition | accuracy: 78.3 -> 81.8 -> 82.2 -> 82.4 | Helped |
| Increase batch size from 24 to 96 (ablation) | ProViCo · UCF101 action recognition | accuracy: 79.9 -> 82.1 -> 84.5 -> 86.1 | Helped |
| Use match probability instead of cosine similarity (ablation) | ProViCo video retrieval · UCF101 video retrieval | R@1: 63.1 -> 64.7 | Helped |
Closest papers by meaning
- Modeling Video As Stochastic Processes for Fine-Grained Video Representation Learning (CVPR 2023 · similarity 0.90)
- Expectation-Maximization Contrastive Learning for Compact Video-and-Language Representations (NeurIPS 2022 · similarity 0.88)
- Frame-Wise Action Representations for Long Videos via Sequence Contrastive Learning (CVPR 2022 · similarity 0.88)
- Spatiotemporal Contrastive Video Representation Learning (CVPR 2021 · similarity 0.87)
- On Discriminative Probabilistic Modeling for Self-Supervised Representation Learning (ICLR 2025 · similarity 0.87)
- Visual Representation Learning with Stochastic Frame Prediction (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.