Research map / Video / Action recognition
Overcoming the Domain Gap in Neural Action Representations
Semih Günel, Florian Aymanns, Sina Honari, Pavan P Ramdya, Pascal Fua
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 swapping augmentation (ablation) | contrastive neural-behavior model · MC2P | across-subject accuracy: + 7.9 | Improved |
| add swapping augmentation (ablation) | contrastive neural-behavior model · MC2P | identity recognition accuracy: -63.0 | Improved |
| add swapping augmentation (ablation) | contrastive neural-behavior model · MC2P | single-subject accuracy: + 2.7 | Improved |
| add calcium augmentation on top of swapping (ablation) | Ours · MC2P | across-subject accuracy: + 2.7 | Improved |
| add calcium augmentation on top of swapping (ablation) | Ours · MC2P | identity recognition accuracy: +1.2 | No clear effect |
| add mix augmentation on top of swapping and calcium (ablation) | Ours · MC2P | across-subject accuracy: + 1.2 | Improved |
| add mix augmentation on top of swapping and calcium (ablation) | Ours · MC2P | identity recognition accuracy: -0.8 | No clear effect |
| add MMD or GRL domain adaptation to SimCLR | SimCLR · MC2P | across-subject accuracy (100%): 50.6 -> 53.1 (MMD), 50.6 -> 52.3 (GRL) | No clear effect |
| add swapping augmentation to SimCLR on pose data | SimCLR · H3.6M Walking, Sitting, Posing | across-subject accuracy (100%): 55.7 -> 73.6 | Improved |
| add swapping augmentation to SimCLR on pose data | SimCLR · ECoG Moving vs Rest | across-subject accuracy (100%): 72.1 -> 81.9 | Improved |
| use neural-only contrastive pretraining instead of multimodal SimCLR | Ours (Neural Only) · MC2P | across-subject accuracy (100%): 50.6 -> 30.6 | Hurt |
| add domain adaptation (MMD or GRL) to SimCLR | SimCLR · MC2P | identity recognition accuracy (100%): 80.3 -> 21.2 (MMD), 80.3 -> 19.1 (GRL) | Improved |
Closest papers by meaning
- MABe22: A Multi-Species Multi-Task Benchmark for Learned Representations of Behavior (ICML 2023 · similarity 0.87)
- Shared-AE: Automatic Identification of Shared Subspaces in High-dimensional Neural and Behavioral Activity (ICLR 2025 · similarity 0.86)
- Neural Encoding and Decoding at Scale (ICML 2025 · similarity 0.86)
- BEAST3D: animal behavioral analysis and neural encoding from multi-view video via Gaussian splatting (NeurIPS 2026 · similarity 0.85)
- Inpainting the Neural Picture: Inferring Unrecorded Brain Area Dynamics from Multi-Animal Datasets (NeurIPS 2025 · similarity 0.85)
- Pose Splatter: A 3D Gaussian Splatting Model for Quantifying Animal Pose and Appearance (NeurIPS 2025 · 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.
