Planning to avoid ambiguous states through Gaussian approximations to non-linear sensors in active inference agents
Fuente:
arXiv
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909319342063616 |
|---|---|
| author | Kouw, Wouter M. |
| author_facet | Kouw, Wouter M. |
| contents | In nature, active inference agents must learn how observations of the world represent the state of the agent. In engineering, the physics behind sensors is often known reasonably accurately and measurement functions can be incorporated into generative models. When a measurement function is non-linear, the transformed variable is typically approximated with a Gaussian distribution to ensure tractable inference. We show that Gaussian approximations that are sensitive to the curvature of the measurement function, such as a second-order Taylor approximation, produce a state-dependent ambiguity term. This induces a preference over states, based on how accurately the state can be inferred from the observation. We demonstrate this preference with a robot navigation experiment where agents plan trajectories. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_01974 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Planning to avoid ambiguous states through Gaussian approximations to non-linear sensors in active inference agents Kouw, Wouter M. Systems and Control Artificial Intelligence Robotics Machine Learning In nature, active inference agents must learn how observations of the world represent the state of the agent. In engineering, the physics behind sensors is often known reasonably accurately and measurement functions can be incorporated into generative models. When a measurement function is non-linear, the transformed variable is typically approximated with a Gaussian distribution to ensure tractable inference. We show that Gaussian approximations that are sensitive to the curvature of the measurement function, such as a second-order Taylor approximation, produce a state-dependent ambiguity term. This induces a preference over states, based on how accurately the state can be inferred from the observation. We demonstrate this preference with a robot navigation experiment where agents plan trajectories. |
| title | Planning to avoid ambiguous states through Gaussian approximations to non-linear sensors in active inference agents |
| topic | Systems and Control Artificial Intelligence Robotics Machine Learning |
| url | https://arxiv.org/abs/2409.01974 |