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
Similar articles
-
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
-
The Neurobiology of Mammalian Navigation.Curr Biol. 2018 Sep 10;28(17):R1023-R1042. doi: 10.1016/j.cub.2018.05.050. Curr Biol. 2018. PMID: 30205053 Review.
-
A unified hierarchical oscillatory network model of head direction cells, spatially periodic cells, and place cells.Eur J Neurosci. 2018 May;47(10):1266-1281. doi: 10.1111/ejn.13918. Epub 2018 Apr 16. Eur J Neurosci. 2018. PMID: 29575125
-
How do spatial learning and memory occur in the brain? Coordinated learning of entorhinal grid cells and hippocampal place cells.J Cogn Neurosci. 2012 May;24(5):1031-54. doi: 10.1162/jocn_a_00200. Epub 2012 Jan 30. J Cogn Neurosci. 2012. PMID: 22288394
-
Framing the grid: effect of boundaries on grid cells and navigation.J Physiol. 2016 Nov 15;594(22):6489-6499. doi: 10.1113/JP270607. Epub 2016 May 10. J Physiol. 2016. PMID: 26969452 Free PMC article. Review.
References
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources
Other Literature Sources