Research map / Video / Video-language understanding
Interactive Episodic Memory with User Feedback
Nikesh Subedi, Loris Bazzani, Ziad Al-Halah
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 FALM with all supervision types (ablation) | GroundNLQ · Ego4D-QnF ablation subset (IoU 0.3) | R1: 29.56 -> 33.13 | Improved |
| add the EM adapter between FALM and the EM-NLQ model (ablation) | ReFocus(GroundNLQ) · Ego4D-QnF ablation subset (IoU 0.3) | R1: 32.46 -> 33.13 | Improved |
| train FALM with temporal supervision only instead of all three (ablation) | FALM · Ego4D-QnF ablation subset (IoU 0.3) | R1: 29.56 -> 32.29 | Improved |
| train FALM with contains supervision only (ablation) | FALM · Ego4D-QnF ablation subset (IoU 0.3) | R1: 29.56 -> 31.08 | Improved |
| train FALM with not-contains supervision only (ablation) | FALM · Ego4D-QnF ablation subset (IoU 0.3) | R1: 29.56 -> 30.89 | Improved |
| combine all three supervision types instead of one (ablation) | FALM · Ego4D-QnF ablation subset (IoU 0.3) | R1: 32.29 -> 33.13 | Improved |
| add FALM to GroundNLQ | GroundNLQ · GoalStep-QnF (IoU 0.3) | R1: 23.3 -> 26.8 | Improved |
| add FALM to OSGNet | OSGNet · Ego4D-QnF (IoU 0.3) | R1: 29.6 -> 32.5 | Improved |
| add FALM to GroundNLQ, zero-shot transfer from Ego4D-QnF | GroundNLQ · HD-EPIC-QnF zero-shot (IoU 0.3) | R1: 6.6 -> 8.2 | Improved |
| use feedback with zero-shot UniTime | UniTime (ZS) · Ego4D-QnF (IoU 0.3) | R1: -5.2 | Hurt |
| finetune UniTime on QnF data | UniTime (FT) · Ego4D-QnF (IoU 0.3) | R1: 21.7 vs 19.9 | No clear effect |
| train GroundNLQ on EM-QnF data without FALM | GroundNLQ · all QnF datasets | R1/R5: Delta <= 1% | No clear effect |
Closest papers by meaning
- Human-inspired Episodic Memory for Infinite Context LLMs (ICLR 2025 · similarity 0.87)
- REMem: Reasoning with Episodic Memory in Language Agent (ICLR 2026 · similarity 0.87)
- Episodic Memories Generation and Evaluation Benchmark for Large Language Models (ICLR 2025 · similarity 0.87)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.86)
- DRESS: Instructing Large Vision-Language Models to Align and Interact with Humans via Natural Language Feedback (CVPR 2024 · similarity 0.86)
- On the Consistency of Video Large Language Models in Temporal Comprehension (CVPR 2025 · 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.
