EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Active sensing with predictive coding and uncertainty minimization

Abdelrahman Sharafeldin, Nabil Imam, Hannah Choi

Patterns · 2024-06 · 5 (6) · 100983

DOI: 10.1016/j.patter.2024.100983 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Patterns, 2024-06. Verfasst von Abdelrahman Sharafeldin und 2 weiteren Autor:innen. Beteiligte Einrichtung: ML@GT, Georgia Institute of Technology, Atlanta, GA 30332, USA. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Science Foundation, Alfred P Sloan Foundation.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We present an end-to-end architecture for embodied exploration inspired by two biological computations: predictive coding and uncertainty minimization. The architecture can be applied to any exploration setting in a task-independent and intrinsically driven manner. We first demonstrate our approach in a maze navigation task and show that it can discover the underlying transition distributions and spatial features of the environment. Second, we apply our model to a more complex active vision task, whereby an agent actively samples its visual environment to gather information. We show that our model builds unsupervised representations through exploration that allow it to efficiently categorize visual scenes. We further show that using these representations for downstream classification leads to superior data efficiency and learning speed compared to other baselines while maintaining lower parameter complexity. Finally, the modular structure of our model facilitates interpretability, allowing us to probe its internal mechanisms and representations during exploration.

Schlagwörter

Stichwörter der Autor:innen

active visionembodied explorationgenerative modelinformation maximizationintrinsic motivationneuro-inspired AIpredictive codingvariational inference

Autor:innen und Institutionen

  • ML@GT, Georgia Institute of Technology, Atlanta, GA 30332, USA.
  • School of Computational Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA.
  • School of Mathematics, Georgia Institute of Technology, Atlanta, GA 30332, USA.

Förderung

  • National Science Foundation
  • Alfred P Sloan Foundation

Zitieren

APA

Sharafeldin, A., Imam, N., & Choi, H. (2024). Active sensing with predictive coding and uncertainty minimization. Patterns, 5(6), 100983. https://doi.org/10.1016/j.patter.2024.100983

Vancouver

Sharafeldin A, Imam N, Choi H. Active sensing with predictive coding and uncertainty minimization. Patterns. 2024;5(6):100983. doi:10.1016/j.patter.2024.100983

BibTeX
@article{Sharafeldin2024,
  author = {Sharafeldin, Abdelrahman and Imam, Nabil and Choi, Hannah},
  title = {Active sensing with predictive coding and uncertainty minimization},
  journal = {Patterns},
  year = {2024},
  volume = {5},
  number = {6},
  pages = {100983},
  doi = {10.1016/j.patter.2024.100983}
}
Kostenlos starten →