Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Combining Recurrent & Bayesian Models for Action Anticipation with Multiple Cues

Mariia Zimokha¹, Lorenzo Jamone, Iran R Roman¹; ¹Queen Mary, University of London

Presenter: Mariia Zimokha

Understanding human action prediction requires modeling rapid temporal integration and deliberative reasoning. Inspired by human studies on multisensory cues, we propose a dual-process model with Reservoir Computing (RC) for temporal processing and Bayesian Networks (BN) for uncertainty-aware probabilistic decisions. The RC integrates sensory cues while the BN processes the output RC states to refine predictions. We tested this integrated framework using simulated reaching tasks with cues such as gaze direction, hand movement, and hand shape. The results indicate that our combined system replicates key aspects of human behavior.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF