Deep hybrid models: infer and plan in a dynamic world
Not on the map. Why?
Abstract (arXiv)
To determine an optimal plan for complex tasks, one often deals with dynamic and hierarchical relationships between several entities. Traditionally, such problems are tackled with optimal control, which relies on the optimization of cost functions; instead, a recent biologically-motivated proposal casts planning and control as an inference process. Active inference assumes that action and perception are two complementary aspects of life whereby the role of the former is to fulfill the predictions inferred by the latter. Here, we present an active inference approach that exploits discrete and continuous processing, based on three features: the representation of potential body configurations in relation to the objects of interest; the use of hierarchical relationships that enable the agent to easily interpret and flexibly expand its body schema for tool use; the definition of potential trajectories related to the agent's intentions, used to infer and plan with dynamic elements at different temporal scales. We evaluate this deep hybrid model on a habitual task: reaching a moving object after having picked a moving tool. We show that the model can tackle the presented task under different conditions. This study extends past work on planning as inference and advances an alternative direction to optimal control.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 28 of its 94 references that are in the map. First posted on arXiv 1 Feb 2024.
Builds on
28 of 94 references.
References · 28
- 2020Active inference on discrete state-spaces: a synthesiscited by 91
- 2020Predictive Coding Approximates Backprop along Arbitrary Computation Graphscited by 89
- 2020Deep active inference agents using Monte-Carlo methodscited by 72
- 2020Sophisticated Inferencecited by 56
- 2021Temporal Predictive Coding For Model-Based Planning In Latent Spacecited by 54
- 2019Deep Active Inference as Variational Policy Gradientscited by 52
- 2022Predictive Coding: Towards a Future of Deep Learning beyond Backpropagation?cited by 45
- 2017Deep Active Inferencecited by 44
- 2020The Neural Coding Framework for Learning Generative Modelscited by 44
- 2020On the Relationship Between Active Inference and Control as Inferencecited by 42
- 2020Model-based Reinforcement Learning: A Surveycited by 39
- 2023Brain-inspired Computational Intelligence via Predictive Codingcited by 30
16 more
- 2022The free energy principle made simpler but not too simplecited by 27
- 2019End-to-End Pixel-Based Deep Active Inference for Body Perception and Actioncited by 24
- 2018Bayesian model reductioncited by 19
- 2021Predictive Coding Can Do Exact Backpropagation on Convolutional and Recurrent Neural Networkscited by 14
- 2023Sequential Memory with Temporal Predictive Codingcited by 14
- 2022Active Predictive Coding: A Unified Neural Framework for Learning Hierarchical World Models for Perception and Planningcited by 10
- 2023Hierarchical generative modelling for autonomous robotscited by 9
- 2020A deep active inference model of the rubber-hand illusioncited by 7
- 2022Object-based active inferencecited by 7
- 2022Disentangling Shape and Pose for Object-Centric Deep Active Inference Modelscited by 6
- 2022Active Predicting Coding: Brain-Inspired Reinforcement Learning for Sparse Reward Robotic Control Problemscited by 6
- 2023Designing explainable artificial intelligence with active inference: A framework for transparent introspection and decision-makingcited by 6
- 2023Deconstructing deep active inferencecited by 5
- 2022Recursive Neural Programs: Variational Learning of Image Grammars and Part-Whole Hierarchiescited by 2
- 2023Object-Centric Scene Representations using Active Inferencecited by 2
- 2023Modeling motor control in continuous-time Active Inference: a surveycited by 2
Built on by
Not cited yet.