| Paper: | PS-1A.15 |
| Session: | Poster Session 1A |
| Location: | Symphony/Overture |
| Session Time: | Thursday, September 6, 16:30 - 18:30 |
| Presentation Time: | Thursday, September 6, 16:30 - 18:30 |
| Presentation: |
Poster
|
| Publication: |
2018 Conference on Cognitive Computational Neuroscience, 5-8 September 2018, Philadelphia, Pennsylvania |
| Paper Title: |
Deep Predictive Coding Models of Sensory Information Processing in the Brain |
| Manuscript: |
Click here to view manuscript |
| DOI: |
https://doi.org/10.32470/CCN.2018.1110-0 |
| Authors: |
Shirin Dora, Cyriel Pennartz, University of Amsterdam, Netherlands; Sander Bohte, Centrum Wiskunde & Informatica, Netherlands |
| Abstract: |
Predictive coding describes how feedforward and feedback connections in the brain enable efficient processing of sensory information in the brain. In this framework, the bottom-up information represents the information received from the external environment and top-down information influences the processing of bottom-up information based upon context, experience, etc. Here, we used predictive coding to construct a deep neural network model of visual information processing in the sensory areas. The network uses an architecture in which each neuron responds only to information in its receptive field. The trained model is used to infer sets of hierarchical causes for real-world images. Here, we show that the model can capture the statistical regularities of real-world images by using the trained model to infer causes behind natural images that are never before presented to the network. |
