Research map / Learning theory and algorithms / Submodular and game algorithms
Unsupervised Procedure Learning via Joint Dynamic Summarization
Ehsan Elhamifar, Zwe Naing
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add state transition dynamics (beta > 0) to summarization (ablation) | JointSeqFL · Inria (K=10) | F1 score %: 34.7 -> 37.3 | Improved |
| add state transition dynamics (beta > 0) to summarization (ablation) | JointSeqFL · ProceL (K=10) | F1 score %: 27.0 -> 28.3 | Improved |
| use dynamic model (SeqFL) instead of independent data points (FL) (ablation) | FL · Inria (K=10) | F1 score %: 27.6 -> 34.3 | Improved |
| summarize videos jointly instead of independently (ablation) | SeqFL · Inria (K=10) | F1 score %: 34.3 -> 37.3 | Improved |
| summarize videos jointly instead of independently (ablation) | SeqFL · Inria (K=7) | F1 score %: 32.6 -> 38.3 | Improved |
| use summarization-based JointSeqFL instead of Alayrac et al. | Alayrac et al. · Inria 'perform CPR' | F1 score %: 31.3 -> 62.9 | Improved |
| use summarization-based JointSeqFL instead of Alayrac et al. | Alayrac et al. · ProceL 'perform CPR' | F1 score %: 17.2 -> 42.1 | Improved |
| increase procedure length K from 7 to 15 | JointSeqFL · Inria | F1 score %: 38.3 -> 38.3 | No clear effect |
| increase beta (dynamic potential weight) to a large value (ablation) | JointSeqFL · Inria/ProceL tasks coffee, Chromecast, smoked salmon | F1 improvement | Hurt |
| increase beta from zero to moderate values (ablation) | JointSeqFL · Inria 'perform CPR' and ProceL 'replace toilet' | F1 improvement | Improved |
| use speech-first Alayrac et al. instead of summarization on 'change tire' | JointSeqFL · Inria 'change tire' (K=10, K=12) | F1 score % | Mixed |
Closest papers by meaning
- Self-Supervised Multi-Task Procedure Learning from Instructional Videos (ECCV 2020 · similarity 0.90)
- Unsupervised Learning From Narrated Instruction Videos (CVPR 2016 · similarity 0.89)
- Collaborative Weakly Supervised Video Correlation Learning for Procedure-Aware Instructional Video Analysis (AAAI 2024 · similarity 0.87)
- Procedure Knowledge Decoupled Distillation Strategy for Procedure Planning in Instructional Videos (AAAI 2025 · similarity 0.87)
- Masked Temporal Interpolation Diffusion for Procedure Planning in Instructional Videos (ICLR 2025 · similarity 0.86)
- SCHEMA: State CHangEs MAtter for Procedure Planning in Instructional Videos (ICLR 2024 · similarity 0.86)
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.
