The author details how their AI agent, with an Active Inference perception pipeline, learned a correction rule that led to over-correction, causing it to misclassify human speech. This incident highlights the challenge of building robust regulation mechanisms in AI systems to prevent over-generalization and suggests a need for more metabolic control.
This is Chapter 7 of 'The Learn Arc' series, delving into Active Inference, POMDPs, sophisticated planning, and Dirichlet learning. It significantly advances the understanding of hierarchical agents and belief-propagated tree search, making it a pivotal part of the series.
This study presents a computational model, based on active inference and a "self-prior" mechanism, where mirror self-recognition behavior emerges spontaneously without external reward. A simulated infant, using only vision and proprioception, successfully removed a sticker from its own face in 70% of cases, demonstrating the self-prior acts as an internal criterion for distinguishing familiar from novel experiences.
This content introduces continuous time in Active Inference, highlighting that an agent's posterior belief about the world's state moves with its dynamics, integrating predictive coding and control theory. It explains how this involves gradient descent in generalized coordinates, encompassing position, velocity, and acceleration, for a unified system.
This article, part of 'The Learn Arc' series, outlines a six-step template to design and ship your first Active Inference agent. It shifts from theory to practical application, guiding users to define the agent's hidden states, observations, and actions.
This article, part of 'The Learn Arc' series on Active Inference, details how abstract mathematical concepts translate into actual neural circuits. It posits the cortex functions as a physical factor graph, where cortical columns are factor nodes and white-matter fibres act as messages in a message passing computation.