DERIN-EPT: Embodied Predictive Thermodynamic Architecture for Biologically-Motivated Edge AI
Abstract
DERIN-EPT (Digital Experiential Reasoning Intelligence Network—Embodied Predictive Thermodynamic) is a novel cognitive architecture implementing the Free Energy Principle and Predictive Processing on edge hardware (NVIDIA Jetson AGX Thor, 128GB unified memory). The architecture introduces 13 innovations in two layers: Core Cognitive Layer: (1) Hardware-as-Body perception, (2) 5-Drive autonomous decision-making, (3) Gut Feeling system, (4) Emergence tracking, (5) World Model, (6) Future Imagination, (7) Self-Evolving Personality, (8) 6-Layer Hierarchical Brain. EPT Innovation Layer: (9) Temporal Eigenstate Memory (TEM), (10) Thermodynamic Compute Layer (TCL), (11) Predictive Attention Manifold (PAM), (12) Self-Modifying Organ Plasticity (SMOP), (13) Embodied Prior Field (EPF). System: 276 core modules, 190,348 lines of code, 969 passing tests. Note: Current implementation developed on Windows. Full deployment on NVIDIA Jetson AGX Thor pending hardware availability.
Fields of Study & Subjects
Cognitive Computing and Networks
Embodied and Extended Cognition
Psychiatry
Mental Health
Neuroscience
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 7 January 2026
- Document type
- Preprint