What question did this study set out to answer?

To develop a framework for modeling emotional dynamics as dynamical systems using physiological signals and world models.

November 14, 2025Open Access

Learning Emotional Dynamics: A World Model Approach

View Full Paper

Ask AI

Bookmark

Authors

Framework demonstrates error reduction in emotion forecasting with physiological signals, suggesting structured emotional dynamics.

To develop a framework for modeling emotional dynamics as dynamical systems using physiological signals and world models.
Utilized a novel framework for emotion modeling based on reinforcement learning and transformer dynamics
Examined three conditioning strategies with the WESAD dataset to assess their impact on prediction accuracy
Implemented multi-step rollout analysis to predict emotional states over time.
Achieved 52.5% error reduction using physiological signals compared to random baseline
Outperformed emotion label conditioning by 34.1%
Demonstrated accurate predictions up to 5 timesteps ahead (R2 = 0.74)
Balanced classification accuracy (76% F1-score) with reconstruction fidelity (MSE=6.7 × 10−3).

Tian ZhiGang (2025) studied this question.

View Full Paper

Ask AI

Bookmark

November 14, 2025Open Access

View Full Paper

Ask AI

Bookmark

Authors

Framework demonstrates error reduction in emotion forecasting with physiological signals, suggesting structured emotional dynamics.

To develop a framework for modeling emotional dynamics as dynamical systems using physiological signals and world models.
Utilized a novel framework for emotion modeling based on reinforcement learning and transformer dynamics
Examined three conditioning strategies with the WESAD dataset to assess their impact on prediction accuracy
Implemented multi-step rollout analysis to predict emotional states over time.
Achieved 52.5% error reduction using physiological signals compared to random baseline
Outperformed emotion label conditioning by 34.1%
Demonstrated accurate predictions up to 5 timesteps ahead (R2 = 0.74)
Balanced classification accuracy (76% F1-score) with reconstruction fidelity (MSE=6.7 × 10−3).

Tian ZhiGang (2025) studied this question.

View Full Paper

Ask AI

Bookmark