Research map / Probabilistic and scientific modeling / Molecules and proteins
MGF: Mixed Gaussian Flow for Diverse Trajectory Prediction
Jiahe Chen, Jinkun Cao, Dahua Lin, Kris M. Kitani, Jiangmiao Pang
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes Mixed Gaussian Flow (MGF), a normalizing flow for trajectory prediction that uses a learned mixed-Gaussian prior instead of a standard Gaussian to improve prediction diversity.
Problem
Normalizing flows with a standard Gaussian prior produce low-diversity trajectory predictions, because the symmetric unimodal prior and supervision losses conflict with the asymmetric, multi-modal distribution of likely future outcomes.
Approach
Replace the standard Gaussian prior of a normalizing flow with a mixed Gaussian prior so that the latent distribution can match asymmetric, multi-modal future trajectory distributions, improving diversity while keeping accuracy.
Experimental setup
Builds on: normalizing flow CVAE
Closest papers by meaning
- MoFlow: One-Step Flow Matching for Human Trajectory Forecasting via Implicit Maximum Likelihood Estimation based Distillation (CVPR 2025 · similarity 0.85)
- Improving Transferability for Cross-Domain Trajectory Prediction via Neural Stochastic Differential Equation (AAAI 2024 · similarity 0.83)
- Normalizing Trajectory Models (NeurIPS 2026 · similarity 0.83)
- SingularTrajectory: Universal Trajectory Predictor Using Diffusion Model (CVPR 2024 · similarity 0.83)
- Flow Matching with Gaussian Process Priors for Probabilistic Time Series Forecasting (ICLR 2025 · similarity 0.83)
- Markovian Flow Matching: Accelerating MCMC with Continuous Normalizing Flows (NeurIPS 2024 · similarity 0.83)
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.
