free energy principle
PulseAugur coverage of free energy principle — every cluster mentioning free energy principle across labs, papers, and developer communities, ranked by signal.
-
New model frames spatial hearing using free energy principle
Researchers have developed a hierarchical generative model to frame spatial hearing within the free energy principle (FEP). This model characterizes how listeners' beliefs about sound direction evolve based on visual fe…
-
AGI Systems Directorate builds Project Lumina AI from scratch, integrating biological mechanisms
AGI Systems Directorate is developing Project Lumina, an autonomous AI agent and large language model built entirely from scratch. The project aims to move beyond transformer architectures by integrating over 500 biolog…
-
Embodied Cognition Meets Extended Mind Thesis in New AI Consciousness Theory
An essay by Dr. Brent Allen Jensen explores the intersection of embodied cognition and the extended mind thesis, proposing that integrated information is not merely data but a form of causal agency. This concept, linked…
-
New framework and benchmark advance robust visual navigation
Researchers have developed FEP-Nav, a novel framework inspired by the Free Energy Principle for real-time adaptation in visual navigation systems. This approach aims to improve robot navigation by minimizing prediction …
-
Neuroscience-inspired self-supervised learning framework introduced
Researchers have introduced Meta-Representational Predictive Coding (MPC), a novel self-supervised learning framework inspired by neuroscience. This approach aims to overcome the limitations of traditional backpropagati…
-
LLM personalization framework models user states for better conversation
Researchers have developed a new framework called PUMA (Prospective User-state Modeling for Action selection) to enhance personalization in large language model conversations. PUMA utilizes the Free Energy Principle to …
-
AI agents need persistent internal states, not just on-demand processing
The current paradigm for AI agents, where they function like smart calculators that activate only when prompted, is a significant limitation. True AI agents require a persistent internal state and continuous operation, …
-
New framework improves U-statistics with active inference for costly labels
Researchers have developed a new active inference framework for U-statistics, aiming to improve estimation efficiency when labeling data is expensive. This approach selectively queries informative labels within a fixed …
-
Category theory framework proposed for defining and comparing AGI architectures
This working paper proposes a formal framework for comparing different Artificial General Intelligence (AGI) architectures using category theory. The authors aim to provide a unified foundation for AGI systems, integrat…
-
New principle unifies Bayesian inference, game theory, and thermodynamics
A new paper introduces the Game-Theoretic Free Energy Principle, a framework that unifies Bayesian inference, game theory, and thermodynamics. This principle suggests that multi-agent systems minimizing local free energ…
-
New MoE Architectures Enhance Efficiency and Performance
Researchers are developing advanced techniques to improve Mixture-of-Experts (MoE) models, particularly addressing challenges in domain transitions and inference efficiency. One approach, inspired by the Free Energy Pri…