Research map / Label-efficient and robust learning / Self-supervised and contrastive

BAL: Bidirectional Autoregression in Latent Space for Learning Human Movement Representations

NeurIPS 2026 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes BAL, a self-supervised method that learns human movement representations from 3D pose sequences via bidirectional autoregression in latent space, avoiding hand-tuned augmentation or masking hyperparameters.

Problem

Self-supervised learning of semantically meaningful human movement representations from 3D pose sequences; existing methods rely on artificial assumptions (augmentation invariance ranges, masked-region recovery at fixed temporal scales) that add sensitive hyperparameters.

Approach

Replaces augmentation-invariance and masked-recovery supervision with bidirectional autoregressive prediction in latent space, so supervision comes from temporal structure of movement itself rather than hand-designed hyperparameter-dependent assumptions.

Closest papers by meaning

• FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · similarity 0.86)
• Poly-Autoregressive Prediction for Modeling Interactions (CVPR 2025 · similarity 0.86)
• AutoTraces: Autoregressive Trajectory Forecasting via Multimodal Large Language Models (CVPR 2026 · similarity 0.86)
• AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.86)
• Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.86)
• Social-Transmotion: Promptable Human Trajectory Prediction (ICLR 2024 · similarity 0.86)