Persistent-Transient Duality in Human Behavior Modeling
This addresses the challenge of accurately predicting human motion in interactive scenarios, which is incremental as it builds on existing neural network approaches.
The paper tackled the problem of modeling persistent and transient aspects of human behavior for motion prediction, and the result was a neural network framework that achieved superior performance in human-object interaction tasks.
We propose to model the persistent-transient duality in human behavior using a parent-child multi-channel neural network, which features a parent persistent channel that manages the global dynamics and children transient channels that are initiated and terminated on-demand to handle detailed interactive actions. The short-lived transient sessions are managed by a proposed Transient Switch. The neural framework is trained to discover the structure of the duality automatically. Our model shows superior performances in human-object interaction motion prediction.