Kickback Cuts Backprop's Red-Tape: Biologically Plausible Credit Assignment in Neural Networks
Open accessThe DOI matches its Crossref record; 8 of the 8 references it cites were found in a registry or an index.
Abstract
Error backpropagation is an extremely effective algorithm for assigning credit in artificial neural networks. However, weight updates under Backprop depend on lengthy recursive computations and require separate output and error messages — features not shared by biological neurons, that are perhaps unnecessary. In this paper, we revisit Backprop and the credit assignment problem. We first decompose Backprop into a collection of interacting learning algorithms; provide regret bounds on the performance of these sub-algorithms; and factorize Backprop's error signals. Using these results, we derive a new credit assignment algorithm for nonparametric regression, Kickback, that is significantly simpler than Backprop. Finally, we provide a sufficient condition for Kickback to follow error gradients, and show that Kickback matches Backprop's performance on real-world regression benchmarks.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Learning representations by back-propagating errors
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986) · Nature · doi:10.1038/323533a0 · Cited by 31,892
Tononi, G., & Cirelli, C. (2014) · Neuron · doi:10.1016/j.neuron.2013.12.025 · Cited by 2,356
The distinct modes of vision offered by feedforward and recurrent processing
Lamme, V. A. F., & Roelfsema, P. R. (2000) · Trends in Neurosciences · doi:10.1016/s0166-2236(00)01657-x · Cited by 2,338 · cited 1×
“Implementing Backprop requires that neurons produce two distinct signals – outputs and errors – whereas only one has been observed in cortex (Lamme and Roelfsema 2000; Roelfsema and van Ooyen 2005).”
BackgroundWhat is the best multi-stage architecture for object recognition?
Jarrett, K., Kavukcuoglu, K., Ranzato, M. A., & LeCun, Y. (2009) · 2009 IEEE 12th International Conference on Computer Vision · doi:10.1109/iccv.2009.5459469 · Cited by 2,153 · cited 2×
“We restrict to rectifiers because they perform well empirically (Jarrett et al. 2009; Nair and Hinton 2010; Glorot, Bordes, and Bengio 2011; Krizhevsky, Sutskever, and Hinton 2012; Zeiler et al. 2013; Dahl, Sainath, and Hinton 2013; Maas, Hannun, and Ng 2013), are more realistic models of cortical neurons than sigmoid…”
MethodImproving deep neural networks for LVCSR using rectified linear units and dropout
Dahl, G. E., Sainath, T. N., & Hinton, G. E. (2013) · 2013 IEEE International Conference on Acoustics, Speech and Signal Processing · doi:10.1109/icassp.2013.6639346 · Cited by 1,310 · cited 1×
“…empirically (Jarrett et al. 2009; Nair and Hinton 2010; Glorot, Bordes, and Bengio 2011; Krizhevsky, Sutskever, and Hinton 2012; Zeiler et al. 2013; Dahl, Sainath, and Hinton 2013; Maas, Hannun, and Ng 2013), are more realistic models of cortical neurons than sigmoid units (Glorot, Bordes, and…”
BackgroundLearning in spiking neural networks by reinforcement of stochastic synaptic transmission
Seung, H. S. (2003) · Neuron · doi:10.1016/s0896-6273(03)00761-x · Cited by 340 · cited 2×
“More recent approaches include REINFORCE (Williams 1992), the hedonistic neurons in (Seung 2003), and the neurons modeled using online learning in (Hu et al. 2013).”
MethodTwenty-Five Lessons from Computational Neuromodulation
Dayan, P. (2012) · Neuron · doi:10.1016/j.neuron.2012.09.027 · Cited by 190
Cited by
Theories of Error Back-Propagation in the Brain
Whittington, J. C. R., & Bogacz, R. (2019) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2018.12.005 · Cited by 420
“There is a diversity of ideas on how the back-propagation algorithm may be approximated in the brain [31 – 36]; however, we review the principles behind a set of related models [11,13,14,37] that have substantial connections with biological data while closely paralleling the back-propagation…”
BackgroundWhittington, J. C. R., & Bogacz, R. (2017) · Neural Computation · doi:10.1162/neco_a_00949 · Cited by 278
“Balduzzi et al. (2014) showed that efficient learning may be achieved by a network that receives a global error signal and in which synaptic weight modification depends jointly on the error and the terms describing the influence of each neuron of final error.”
BackgroundToward an integration of deep learning and neuroscience
Marblestone, A. H., Wayne, G., & Kording, K. P. (2016) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2016.00094 · Cited by 119
Metadata
- Authors
- Balduzzi, D., Vanchinathan, H., & Buhmann, J.
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2015-02-10
- Volume
- 29
- Issue
- 1
- ISSN
- 2374-3468
- 2159-5399
- OpenAlex
- W2949933418
- Cited by
- 29
- References
- 8
- FWCI
- 2.50
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.