The Limits of Feedforward Vision: Recurrent Processing Promotes Robust Object Recognition when Objects Are Degraded
By Dean Wyatte, Tim Curran, Randall C. O’Reilly. 2012
124 citations recorded by the source corpus.
Subfields: Cognitive Neuroscience
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Cognitive Neuroscience
- J. Urgen Schmidhuber
- Pamela K. Douglas
- Faustino J. Gomez
- Seth A. Herd
- Pulkit Agrawal
- Jitendra Malik
- Dean Wyatte
- David J. Jilk
Citations in the corpus
- Deep learning in neural networks: An overview (2014, neutral): “Compare other recent RNNs for object recognition (Wyatte et al., 2012; OReilly et al., 2013), extending earlier work on biologically plausible learning rules for RNNs (O’Reilly, 1996).”
- Deep Networks with Internal Selective Attention through Feedback Connections (2014, background): “Recurrent connections seem to rely heavily on competitive inhibition and other feedback to make object recognition more robust [38, 39].”
- Human Pose Estimation with Iterative Error Feedback (2015, background): “Neuroscience models of the human brain suggest that feedforward connections act as information carriers while numerous feedback connections act as modulators or competitive inhibitors to aid feature grouping [14], figure-ground segregation [16] and object recognition [46].”
- Cognitive computational neuroscience (2018, background): “However, recurrent neural networks, which reprocess the outputs of their units and generate complex dynamics, have brought additional engineering advances(106) and better capture the recurrent signaling in brains.(107,108,48,109) Whereas feedforward networks are universal function approximators, rec”
- Contextual Priming and Feedback for Faster R-CNN (2016, background): “But so far, our focus has been on designing deeper [5, 6] feedforward architectures, even when there is a broad agreement on the importance of feedback connections [17, 41, 42] and limitations of purely feedforward recognition [43, 44] in human visual systems.”
- Recurrent Processing during Object Recognition (2013, background): “Recent psychophysical studies from our lab that use backward masking to disrupt ongoing recurrent processing are consistent with this prediction (Wyatte et al., 2012a).”