2018

Dendritic cortical microcircuits approximate the backpropagation algorithm

Sacramento, João, Costa, Rui Ponte, Bengio, Yoshua et al.

Understand

Deep learning has seen remarkable developments over the last years, many of them inspired by neuroscience.

However, the main learning mechanism behind these advances - error backpropagation - appears to be at odds with neurobiology.
Here, we introduce a multilayer neuronal network model with simplified dendritic compartments in which error-driven synaptic plasticity adapts the network towards a global desired output.
In contrast to previous work our model does not require separate phases and synaptic learning is driven by local dendritic prediction errors continuously in time.

Built on

A learning algorithm for Boltzmann machines
Ackley, D. H., Hinton, G. E., and Sejnowski, T. J. (1985) · 1985
Earlier work this paper cites.
Learning representations by back-propagating errors
Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986) · 1986
Earlier work this paper cites.
Competitive learning: From interactive activation to adaptive resonance
Grossberg, S. (1987) · 1987
Earlier work this paper cites.
Learning representations by recirculation
Hinton, G. E. and McClelland, J. L. (1988) · 1988
Earlier work this paper cites.
A theoretical framework for back-propagation
LeCun, Y. (1988) · 1988
Earlier work this paper cites.
The recent excitement about neural networks
Crick, F. (1989) · 1989
Earlier work this paper cites.
Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory
McClelland, J. L., McNaughton, B. L., and O’reilly, R. C. (1995) · 1995
Earlier work this paper cites.
Biologically plausible error-driven learning using local activation differences: The generalized recirculation algorithm
O’Reilly, R. C. (1996) · 1996
Earlier work this paper cites.
Online algorithms and stochastic approximations
Bottou, L. (1998) · 1998
Earlier work this paper cites.
Reinforcement learning: An introduction
Sutton, R. S. and Barto, A. G. (1998) · 1998
Earlier work this paper cites.
Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects
Rao, R. P. and Ballard, D. H. (1999) · 1999
Earlier work this paper cites.
Learning with two sites of synaptic integration
Körding, K. P. and König, P. (2000) · 2000
Earlier work this paper cites.
Supervised and unsupervised learning with two sites of synaptic integration
Körding, K. P. and König, P. (2001) · 2001
Earlier work this paper cites.
Rate, Timing, and Cooperativity Jointly Determine Cortical Synaptic Plasticity
Sjöström, P. J., Turrigiano, G. G., and Nelson, S. B. (2001) · 2001
Earlier work this paper cites.
Equivalence of backpropagation and contrastive Hebbian learning in a layered network
Xie, X. and Seung, H. S. (2003) · 2003
Earlier work this paper cites.
A theory of cortical responses
Friston, K. (2005) · 2005
Earlier work this paper cites.
Attention-gated reinforcement learning of internal representations for classification
Roelfsema, P. R. and van Ooyen, A. (2005) · 2005
Earlier work this paper cites.
Unsupervised learning of visual features through spike timing dependent plasticity
Masquelier, T. and Thorpe, S. (2007) · 2007
Earlier work this paper cites.
Pyramidal neurons: dendritic structure and synaptic integration
Spruston, N. (2008) · 2008
Earlier work this paper cites.
The subcellular organization of neocortical excitatory connections
Petreanu, L., Mao, T., Sternson, S. M., and Svoboda, K. (2009) · 2009
Earlier work this paper cites.

Similar

Connectivity reflects coding: a model of voltage-based stdp with homeostasis
Clopath, C., Büsing, L., Vasilaki, E., and Gerstner, W. (2010) · 2010
Cited alongside, same era.
Developmental sensory experience balances cortical excitation and inhibition
Dorrn, A. L., Yuan, K., Barker, A. J., Schreiner, C. E., and Froemke, R. C. (2010) · 2010
Cited alongside, same era.
Spatio-temporal credit assignment in neuronal population learning
Friedrich, J., Urbanczik, R., and Senn, W. (2011) · 2011
Cited alongside, same era.
Inhibitory plasticity balances excitation and inhibition in sensory pathways and memory networks
Vogels, T. P., Sprekeler, H., Zenke, F., Clopath, C., and Gerstner, W. (2011) · 2011
Cited alongside, same era.
Balancing feed-forward excitation and inhibition via Hebbian inhibitory synaptic plasticity
Luz, Y. and Shamir, M. (2012) · 2012
Cited alongside, same era.
Activity in motor-sensory projections reveals distributed coding in somatosensation
Petreanu, L., Gutnisky, D. A., Huber, D., Xu, N.-l., O’Connor, D. H., Tian, L., Looger, L., and Svoboda, K. (2012) · 2012
Cited alongside, same era.
A cellular mechanism for cortical associations: an organizing principle for the cerebral cortex
Larkum, M. (2013) · 2013
Cited alongside, same era.
Preference distributions of primary motor cortex neurons reflect control solutions optimized for limb biomechanics
Lillicrap, T. P. and Scott, S. H. (2013) · 2013
Cited alongside, same era.
Bayesian computation emerges in generic cortical microcircuits through spike-timing-dependent plasticity
Nessler, B., Pfeiffer, M., Buesing, L., and Maass, W. (2013) · 2013
Cited alongside, same era.
How auto-encoders could provide credit assignment in deep networks via target propagation
Original
Bengio, Y. (2014) · 2014
Cited alongside, same era.
Deep supervised, but not unsupervised, models may explain it cortical representation
Khaligh-Razavi, S.-M. and Kriegeskorte, N. (2014) · 2014
Cited alongside, same era.
Learning by the dendritic prediction of somatic spiking
Urbanczik, R. and Senn, W. (2014) · 2014
Cited alongside, same era.

Then

What learning systems do intelligent agents need? complementary learning systems theory updated
Kumaran, D., Hassabis, D., and McClelland, J. L. (2016) · 2016
Later among the works it cites.
Random synaptic feedback weights support error backpropagation for deep learning
Lillicrap, T. P., Cownden, D., Tweed, D. B., and Akerman, C. J. (2016) · 2016
Later among the works it cites.
Toward an integration of deep learning and neuroscience
Marblestone, A. H., Wayne, G., and Kording, K. P. (2016) · 2016
Later among the works it cites.
Direct feedback alignment provides learning in deep neural networks
Nøkland, A. (2016) · 2016
Later among the works it cites.
Behavioral-state modulation of inhibition is context-dependent and cell type specific in mouse visual cortex
Pakan, J. M., Lowe, S. C., Dylda, E., Keemink, S. W., Currie, S. P., Coutts, C. A., Rochefort, N. L., and Mrsic-Flogel, T. D. (2016) · 2016
Later among the works it cites.
Somatostatin-expressing neurons in cortical networks
Urban-Ciecko, J. and Barth, A. L. (2016) · 2016
Later among the works it cites.
Using goal-driven deep learning models to understand sensory cortex
Yamins, D. L. and DiCarlo, J. J. (2016) · 2016
Later among the works it cites.
Mismatch receptive fields in mouse visual cortex
Zmarz, P. and Keller, G. B. (2016) · 2016
Later among the works it cites.
Visuomotor coupling shapes the functional development of mouse visual cortex
Attinger, A., Wang, B., and Keller, G. B. (2017) · 2017
Later among the works it cites.
Modeling somatic and dendritic spike mediated plasticity at the single neuron and network level
Bono, J. and Clopath, C. (2017) · 2017
Later among the works it cites.
Cortical microcircuits as gated-recurrent neural networks
Costa, R. P., Assael, Y. M., Shillingford, B., de Freitas, N., and Vogels, T. P. (2017) · 2017
Later among the works it cites.
Towards deep learning with segregated dendrites
Guerguiev, J., Lillicrap, T. P., and Richards, B. A. (2017) · 2017
Later among the works it cites.
A Sensorimotor Circuit in Mouse Cortex for Visual Flow Predictions
Leinweber, M., Ward, D. R., Sobczak, J. M., Attinger, A., and Keller, G. B. (2017) · 2017
Later among the works it cites.
Equilibrium propagation: Bridging the gap between energy-based models and backpropagation
Scellier, B. and Bengio, Y. (2017) · 2017
Later among the works it cites.
High-level prediction signals in a low-level area of the macaque face-processing hierarchy
Schwiedrzik, C. M. and Freiwald, W. A. (2017) · 2017
Later among the works it cites.
An approximation of the error backpropagation algorithm in a predictive coding network with local Hebbian synaptic plasticity
Whittington, J. C. R. and Bogacz, R. (2017) · 2017
Later among the works it cites.
Input-specific nmdar-dependent potentiation of dendritic gabaergic inhibition
Chiu, C. Q., Martenson, J. S., Yamazaki, M., Natsume, R., Sakimura, K., Tomita, S., Tavalin, S. J., and Higley, M. J. (2018) · 2018
Closest in time.
A task-optimized neural network replicates human auditory behavior, predicts brain responses, and reveals a cortical processing hierarchy
Kell, A. J., Yamins, D. L., Shook, E. N., Norman-Haignere, S. V., and McDermott, J. H. (2018) · 2018
Closest in time.
Control of synaptic plasticity in deep cortical networks
Roelfsema, P. R. and Holtmaat, A. (2018) · 2018
Closest in time.
Predictive plasticity in dendrites: from a computational principle to experimental data (in preparation)
Spicher, D., Clopath, C., and Senn, W. (2018) · 2018
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…