Trajectory Inference

Trajectory inference aims to reconstruct the continuous evolution of a system from limited, often incomplete, observations at discrete time points. Current research focuses on developing robust algorithms, such as those based on optimal transport, Schrödinger bridges, and mean-field Langevin dynamics, to infer trajectories even with partially observed data or noisy measurements, often incorporating dynamic models and manifold learning techniques to improve accuracy and efficiency. These advancements have significant implications for diverse fields, including autonomous vehicle navigation, single-cell biology analysis, and the study of complex dynamical systems, enabling more accurate modeling and prediction of temporal processes.