Deep Predictive Coding Network for Object Recognition

Haiguang Wen, Kuan Han, Junxing Shi, Yizhen Zhang, Eugenio Culurciello, Zhongming Liu
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:5266-5275, 2018.

Abstract

Based on the predictive coding theory in neuro- science, we designed a bi-directional and recur- rent neural net, namely deep predictive coding networks (PCN), that has feedforward, feedback, and recurrent connections. Feedback connections from a higher layer carry the prediction of its lower-layer representation; feedforward connec- tions carry the prediction errors to its higher-layer. Given image input, PCN runs recursive cycles of bottom-up and top-down computation to update its internal representations and reduce the differ- ence between bottom-up input and top-down pre- diction at every layer. After multiple cycles of recursive updating, the representation is used for image classification. With benchmark datasets (CIFAR-10/100, SVHN, and MNIST), PCN was found to always outperform its feedforward-only counterpart: a model without any mechanism for recurrent dynamics, and its performance tended to improve given more cycles of computation over time. In short, PCN reuses a single architecture to recursively run bottom-up and top-down pro- cesses to refine its representation towards more accurate and definitive object recognition.

Cite this Paper


BibTeX
@InProceedings{pmlr-v80-wen18a, title = {Deep Predictive Coding Network for Object Recognition}, author = {Wen, Haiguang and Han, Kuan and Shi, Junxing and Zhang, Yizhen and Culurciello, Eugenio and Liu, Zhongming}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {5266--5275}, year = {2018}, editor = {Dy, Jennifer and Krause, Andreas}, volume = {80}, series = {Proceedings of Machine Learning Research}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/wen18a/wen18a.pdf}, url = {http://proceedings.mlr.press/v80/wen18a.html}, abstract = {Based on the predictive coding theory in neuro- science, we designed a bi-directional and recur- rent neural net, namely deep predictive coding networks (PCN), that has feedforward, feedback, and recurrent connections. Feedback connections from a higher layer carry the prediction of its lower-layer representation; feedforward connec- tions carry the prediction errors to its higher-layer. Given image input, PCN runs recursive cycles of bottom-up and top-down computation to update its internal representations and reduce the differ- ence between bottom-up input and top-down pre- diction at every layer. After multiple cycles of recursive updating, the representation is used for image classification. With benchmark datasets (CIFAR-10/100, SVHN, and MNIST), PCN was found to always outperform its feedforward-only counterpart: a model without any mechanism for recurrent dynamics, and its performance tended to improve given more cycles of computation over time. In short, PCN reuses a single architecture to recursively run bottom-up and top-down pro- cesses to refine its representation towards more accurate and definitive object recognition.} }
Endnote
%0 Conference Paper %T Deep Predictive Coding Network for Object Recognition %A Haiguang Wen %A Kuan Han %A Junxing Shi %A Yizhen Zhang %A Eugenio Culurciello %A Zhongming Liu %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-wen18a %I PMLR %P 5266--5275 %U http://proceedings.mlr.press/v80/wen18a.html %V 80 %X Based on the predictive coding theory in neuro- science, we designed a bi-directional and recur- rent neural net, namely deep predictive coding networks (PCN), that has feedforward, feedback, and recurrent connections. Feedback connections from a higher layer carry the prediction of its lower-layer representation; feedforward connec- tions carry the prediction errors to its higher-layer. Given image input, PCN runs recursive cycles of bottom-up and top-down computation to update its internal representations and reduce the differ- ence between bottom-up input and top-down pre- diction at every layer. After multiple cycles of recursive updating, the representation is used for image classification. With benchmark datasets (CIFAR-10/100, SVHN, and MNIST), PCN was found to always outperform its feedforward-only counterpart: a model without any mechanism for recurrent dynamics, and its performance tended to improve given more cycles of computation over time. In short, PCN reuses a single architecture to recursively run bottom-up and top-down pro- cesses to refine its representation towards more accurate and definitive object recognition.
APA
Wen, H., Han, K., Shi, J., Zhang, Y., Culurciello, E. & Liu, Z.. (2018). Deep Predictive Coding Network for Object Recognition. Proceedings of the 35th International Conference on Machine Learning, in Proceedings of Machine Learning Research 80:5266-5275 Available from http://proceedings.mlr.press/v80/wen18a.html.

Related Material