Grid-like units help deep learning agent to navigate
- PMID: 29948880
- DOI: 10.3758/s13420-018-0329-y
Grid-like units help deep learning agent to navigate
Abstract
An artificial-intelligence model based on deep learning developed units in a hidden layer that resembled mammalian grid cells in the hippocampus when the agent was taught to integrate paths. The full model performed sophisticated navigational tasks-in some cases even better than a human.
Keywords: Artificial intelligence; Modeling; Navigation; Spatial learning.
Comment on
-
Vector-based navigation using grid-like representations in artificial agents.Nature. 2018 May;557(7705):429-433. doi: 10.1038/s41586-018-0102-6. Epub 2018 May 9. Nature. 2018. PMID: 29743670
References
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources
Other Literature Sources
