Research map / Video / Video segmentation and tracking
Learning Disentangled Behavior Embeddings
Read the paperCheck 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 |
|---|---|---|---|
| add disentanglement (DBE pose embeddings instead of CAE) (ablation) | BehaveNet pose embedding · Hand-reach session classification | session classification accuracy: 100.00 -> 28.21 | Helped |
| add disentanglement (DBE) on WFCI (ablation) | BehaveNet pose embedding · WFCI session classification | session classification accuracy: 99.99 -> 35.49 | Helped |
| add stochastic dynamics to DBE (VDBE) (ablation) | DBE · Hand-reach session classification | session classification accuracy: 28.21 -> 17.94 | Helped |
| replace CAE embeddings with DBE in ARHMM (k=30) (ablation) | BehaveNet · Hand-reach motif segmentation | accuracy: 53.09 -> 71.49 | Helped |
| replace CAE embeddings with DBE in ARHMM (k=15) (ablation) | BehaveNet · Hand-reach motif segmentation | accuracy: 50.52 -> 64.04 | Helped |
| replace DLC trajectories with DBE embeddings as VAME input (k=30) (ablation) | VAME · Hand-reach motif segmentation | accuracy: 57.17 -> 81.25 | Helped |
| replace DLC trajectories with DBE embeddings as VAME input (k=30) (ablation) | VAME · Hand-reach motif segmentation | ARI: 44.09 -> 67.33 | Helped |
| replace DLC trajectories with DBE embeddings as VAME input (k=30) (ablation) | VAME · Hand-reach motif segmentation | NMI: 42.07 -> 67.31 | Helped |
| use VAME temporal model instead of ARHMM with DBE (k=30) (ablation) | DBE · Hand-reach motif segmentation | accuracy: 71.49 -> 81.25 | Helped |
| increase number of motifs from k=15 to k=30 (ablation) | DBE + VAME · Hand-reach motif segmentation | accuracy: 71.31 -> 81.25 | Helped |
| increase number of motifs from k=15 to k=30 (ablation) | BehaveNet · Hand-reach motif segmentation | accuracy: 50.52 -> 53.09 | Helped |
| increase number of motifs from k=15 to k=30 (ablation) | DBE + ARHMM · Hand-reach motif segmentation | NMI: 42.06 -> 48.96 | Helped |
Closest papers by meaning
- BehaveNet: nonlinear embedding and Bayesian neural decoding of behavioral videos (NeurIPS 2019 · similarity 0.90)
- Exploring Behavior-Relevant and Disentangled Neural Dynamics with Generative Diffusion Models (NeurIPS 2024 · similarity 0.86)
- Time-Evolving Dynamical System for Learning Latent Representations of Mouse Visual Neural Activity (NeurIPS 2025 · similarity 0.85)
- Pose Splatter: A 3D Gaussian Splatting Model for Quantifying Animal Pose and Appearance (NeurIPS 2025 · similarity 0.85)
- Agent-to-Sim: Learning Interactive Behavior Models from Casual Longitudinal Videos (ICLR 2025 · similarity 0.84)
- Neural Encoding and Decoding at Scale (ICML 2025 · similarity 0.84)
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.