Minimizing Control for Credit Assignment with Strong Feedback

Alexander Meulemans, Matilde Tristany Farinha, Maria R. Cervera, João Sacramento, Benjamin F. Grewe
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:15458-15483, 2022.

Abstract

The success of deep learning ignited interest in whether the brain learns hierarchical representations using gradient-based learning. However, current biologically plausible methods for gradient-based credit assignment in deep neural networks need infinitesimally small feedback signals, which is problematic in biologically realistic noisy environments and at odds with experimental evidence in neuroscience showing that top-down feedback can significantly influence neural activity. Building upon deep feedback control (DFC), a recently proposed credit assignment method, we combine strong feedback influences on neural activity with gradient-based learning and show that this naturally leads to a novel view on neural network optimization. Instead of gradually changing the network weights towards configurations with low output loss, weight updates gradually minimize the amount of feedback required from a controller that drives the network to the supervised output label. Moreover, we show that the use of strong feedback in DFC allows learning forward and feedback connections simultaneously, using learning rules fully local in space and time. We complement our theoretical results with experiments on standard computer-vision benchmarks, showing competitive performance to backpropagation as well as robustness to noise. Overall, our work presents a fundamentally novel view of learning as control minimization, while sidestepping biologically unrealistic assumptions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-meulemans22a, title = {Minimizing Control for Credit Assignment with Strong Feedback}, author = {Meulemans, Alexander and Farinha, Matilde Tristany and Cervera, Maria R. and Sacramento, Jo{\~a}o and Grewe, Benjamin F.}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {15458--15483}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/meulemans22a/meulemans22a.pdf}, url = {https://proceedings.mlr.press/v162/meulemans22a.html}, abstract = {The success of deep learning ignited interest in whether the brain learns hierarchical representations using gradient-based learning. However, current biologically plausible methods for gradient-based credit assignment in deep neural networks need infinitesimally small feedback signals, which is problematic in biologically realistic noisy environments and at odds with experimental evidence in neuroscience showing that top-down feedback can significantly influence neural activity. Building upon deep feedback control (DFC), a recently proposed credit assignment method, we combine strong feedback influences on neural activity with gradient-based learning and show that this naturally leads to a novel view on neural network optimization. Instead of gradually changing the network weights towards configurations with low output loss, weight updates gradually minimize the amount of feedback required from a controller that drives the network to the supervised output label. Moreover, we show that the use of strong feedback in DFC allows learning forward and feedback connections simultaneously, using learning rules fully local in space and time. We complement our theoretical results with experiments on standard computer-vision benchmarks, showing competitive performance to backpropagation as well as robustness to noise. Overall, our work presents a fundamentally novel view of learning as control minimization, while sidestepping biologically unrealistic assumptions.} }
Endnote
%0 Conference Paper %T Minimizing Control for Credit Assignment with Strong Feedback %A Alexander Meulemans %A Matilde Tristany Farinha %A Maria R. Cervera %A João Sacramento %A Benjamin F. Grewe %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-meulemans22a %I PMLR %P 15458--15483 %U https://proceedings.mlr.press/v162/meulemans22a.html %V 162 %X The success of deep learning ignited interest in whether the brain learns hierarchical representations using gradient-based learning. However, current biologically plausible methods for gradient-based credit assignment in deep neural networks need infinitesimally small feedback signals, which is problematic in biologically realistic noisy environments and at odds with experimental evidence in neuroscience showing that top-down feedback can significantly influence neural activity. Building upon deep feedback control (DFC), a recently proposed credit assignment method, we combine strong feedback influences on neural activity with gradient-based learning and show that this naturally leads to a novel view on neural network optimization. Instead of gradually changing the network weights towards configurations with low output loss, weight updates gradually minimize the amount of feedback required from a controller that drives the network to the supervised output label. Moreover, we show that the use of strong feedback in DFC allows learning forward and feedback connections simultaneously, using learning rules fully local in space and time. We complement our theoretical results with experiments on standard computer-vision benchmarks, showing competitive performance to backpropagation as well as robustness to noise. Overall, our work presents a fundamentally novel view of learning as control minimization, while sidestepping biologically unrealistic assumptions.
APA
Meulemans, A., Farinha, M.T., Cervera, M.R., Sacramento, J. & Grewe, B.F.. (2022). Minimizing Control for Credit Assignment with Strong Feedback. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:15458-15483 Available from https://proceedings.mlr.press/v162/meulemans22a.html.

Related Material