Research map / Video / Action recognition
Learning Recurrent Representations for Hierarchical Behavior Modeling
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 future-motion prediction as an auxiliary task and train on unlabeled sequences (ablation) | BESNet versus BENet · Fly-vs-Fly, FlyBowl, and IAM-OnDB with scarce training labels | frame-wise F1 | Helped |
| Remove the generative motion-prediction part (ablation) | BESNet · Fly-vs-Fly, FlyBowl, and IAM-OnDB with scarce training labels | frame-wise F1 | Hurt |
| Apply flat-filter smoothing to classification outputs (ablation) | BESNet · Fly-vs-Fly | bout-wise performance | Helped |
| Use recurrent cells instead of a non-recurrent model (ablation) | Motion-prediction models · FlyBowl test set | motion-prediction performance | Helped |
| Add diagonal connections between discriminative and generative layers (ablation) | BESNet · FlyBowl test set | motion-prediction performance | Helped |
| Remove diagonal connections (ablation) | BESNet trained without action labels · IAM-OnDB | writer-identity representation | Hurt |
| Represent motion prediction as a distribution over bins rather than deterministic motion (ablation) | SynthFly model · SynthFly simulation | trajectory quality | Helped |
Closest papers by meaning
- Learning interpretable control inputs and dynamics underlying animal locomotion (ICLR 2024 · similarity 0.89)
- Learning Reusable Motor Motifs for Continuous Animal Behavior Modeling (NeurIPS 2026 · similarity 0.88)
- Learning Options for Compositional Motor Control with Adapter Banks (NeurIPS 2026 · similarity 0.85)
- Flow Snapshot Neurons in Action: Deep Neural Networks Generalize to Biological Motion Perception (NeurIPS 2024 · similarity 0.84)
- Humanoid Locomotion as Next Token Prediction (NeurIPS 2024 · similarity 0.84)
- Mechanistic Interpretability of RNNs emulating Hidden Markov Models (NeurIPS 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.