Filters








122 Hits in 4.4 sec

Dendritic error backpropagation in deep cortical microcircuits [article]

João Sacramento and Rui Ponte Costa and Yoshua Bengio and Walter Senn
2017 arXiv   pre-print
We demonstrate the learning capabilities of the model in a number of tasks and show that it approximates the classical error backpropagation algorithm.  ...  Our framework makes several experimental predictions on the function of dendritic integration and cortical microcircuits, is consistent with recent observations of cross-area learning, and suggests a biological  ...  WS thanks Matthew Larkum for many discussions on dendritic processing and the option of dendritic error representation. In addition, JS thanks Elena Kreutzer, Pascal Leimer and Martin T.  ... 
arXiv:1801.00062v1 fatcat:uruwi5mq3bgxdl7fgeokof55dm

Dendritic cortical microcircuits approximate the backpropagation algorithm [article]

João Sacramento, Rui Ponte Costa, Yoshua Bengio, Walter Senn
2018 arXiv   pre-print
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.  ...  Moreover, our framework is consistent with recent observations of learning between brain areas and the architecture of cortical microcircuits.  ...  Supplementary Material: Dendritic cortical microcircuits approximate the backpropagation algorithm The dendritic cortical circuit learns to predict self-generated top-down input Figure S1 : Dendritic  ... 
arXiv:1810.11393v1 fatcat:jetulqdayrdrtnbh4ejzfqtusq

A Computational Framework of Cortical Microcircuits Approximates Sign-concordant Random Backpropagation [article]

Yukun Yang, Peng Li
2022 arXiv   pre-print
In this work, we answer the question of whether it is possible to realize random backpropagation solely based on mechanisms observed in neuroscience.  ...  Comprising three types of cells and two types of synaptic connectivity, the proposed microcircuit architecture computes and propagates error signals through local feedback connections and supports the  ...  Conclusion In this work, we construct a novel framework to implement bio-plausible random error backpropagation.  ... 
arXiv:2205.07292v1 fatcat:ewb5jxuw6bho3fyinfnkdxoydu

Current State and Future Directions for Learning in Biological Recurrent Neural Networks: A Perspective Piece

Roy Henha Eyono, Ellen Boven, Arna Ghosh, Joseph Pemberton, Franz Scherr, Claudia Clopath, Rui Ponte Costa, Wolfgang Maass, Blake A. Richards, Cristina Savin, Katharina Wilmes, Luke Y. Prince
2022 Neurons, Behavior, Data analysis, and Theory  
We conclude with recommendations for both theoretical and experimental neuroscientists when designing new studies that could help bring clarity to these issues. a recurrent neural networks, backpropagation  ...  The key issues discussed in this review include: synaptic plasticity, neural circuits, theory-experiment divide, and objective functions.  ...  Backpropagation through time and the brain. Current opinion in neurobiology 2019;55:82-89. [2] Jozefowicz R, Zaremba W, Sutskever I. An empirical exploration of recurrent network architectures.  ... 
doi:10.51628/001c.35302 fatcat:ahqqn4rpw5dw3mu3n5tzb2auqq

Current State and Future Directions for Learning in Biological Recurrent Neural Networks: A Perspective Piece [article]

Luke Y. Prince, Roy Henha Eyono, Ellen Boven, Arna Ghosh, Joe Pemberton, Franz Scherr, Claudia Clopath, Rui Ponte Costa, Wolfgang Maass, Blake A. Richards, Cristina Savin, Katharina Anna Wilmes
2022 arXiv   pre-print
The key issues discussed in this review include: synaptic plasticity, neural circuits, theory-experiment divide, and objective functions.  ...  We provide a brief review of the common assumptions about biological learning with findings from experimental neuroscience and contrast them with the efficiency of gradient-based learning in recurrent  ...  We would also like to thank all other audience members from around the world who participated on the day in the chat, and everybody else who has watched the replay since it was posted online.  ... 
arXiv:2105.05382v2 fatcat:7vyo56k4rva7vdskumr6zxer4i

A simple normative network approximates local non-Hebbian learning in the cortex [article]

Siavash Golkar, David Lipshutz, Yanis Bahroun, Anirvan M. Sengupta, Dmitri B. Chklovskii
2020 arXiv   pre-print
We detail how, in our model, the calcium plateau potential can be interpreted as a backpropagating error signal.  ...  Neuroscience experiments demonstrate that the processing of sensory inputs by cortical neurons is modulated by instructive signals which provide context and task-relevant information.  ...  • We demonstrate within the confines of our model how the calcium plateau potential in cortical microcircuits encodes a backpropagating error signal. • We show numerically on a real-world dataset that  ... 
arXiv:2010.12660v1 fatcat:ly5hr4mr2zc2tkg2pwddzsowfm

Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons [article]

Paul Haider, Benjamin Ellenberger, Laura Kriener, Jakob Jordan, Walter Senn, Mihai A. Petrovici
2021 arXiv   pre-print
The resulting model can be interpreted as a biologically plausible approximation of error backpropagation in deep cortical networks with continuous-time, leaky neuronal dynamics and continuously active  ...  In hierarchical models of cortical networks each layer thus introduces a response lag.  ...  Within the context of LE, the microcircuit model used to implement error backpropagation carries several implications for cortical phenomenology beyond specific connectivity patterns.  ... 
arXiv:2110.14549v1 fatcat:gbypwhc4rfcjbh5s5qlm7ds67a

Toward an Integration of Deep Learning and Neuroscience

Adam H. Marblestone, Greg Wayne, Konrad P. Kording
2016 Frontiers in Computational Neuroscience  
Here we think about the brain in terms of these ideas.  ...  In machine learning, however, artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively  ...  Some cortical microcircuits 62 It would be interesting to study these questions in specific brain systems.  ... 
doi:10.3389/fncom.2016.00094 pmid:27683554 pmcid:PMC5021692 fatcat:yikwc4h5yvfj7gwzlimtw5n6ai

Towards an integration of deep learning and neuroscience [article]

Adam Marblestone, Greg Wayne, Konrad Kording
2016 arXiv   pre-print
Here we think about the brain in terms of these ideas.  ...  In machine learning, however, artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively  ...  We thank Miles Brundage for an excellent Twitter feed of deep learning papers.  ... 
arXiv:1606.03813v1 fatcat:tmmholydqbcplbc5ihg76yip6e

Towards an integration of deep learning and neuroscience [article]

Adam Henry Marblestone, Greg Wayne, Konrad P Kording
2016 bioRxiv   pre-print
Here we think about the brain in terms of these ideas.  ...  In machine learning, however, artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively  ...  Some cortical microcircuits could then, perhaps, compute the cost functions that are to be delivered to other cortical microcircuits.  ... 
doi:10.1101/058545 fatcat:4ryejpe2tnf7dgoaqhoastoiya

Linking canonical microcircuits and neuronal activity: Dynamic causal modelling of laminar recordings

D.A. Pinotsis, J.P. Geerts, L. Pinto, T.H.B. FitzGerald, V. Litvak, R. Auksztulewicz, K.J. Friston
2017 NeuroImage  
This is particularly interesting from the perspective of predictive coding, where neuromodulators are thought to boost prediction errors that ascend the cortical hierarchy.  ...  ) superficial layer activity relative to deep layers.  ...  To ensure the prior constraints properly accommodate spatiotemporal dynamics within the cortical microcircuit and its neuronal compartments (e.g. delays due to spread of current throughout the dendritic  ... 
doi:10.1016/j.neuroimage.2016.11.041 pmid:27871922 pmcid:PMC5312791 fatcat:vgvd2vfkr5hj7lmpvkq6zwdcvy

Relaxing the Constraints on Predictive Coding Models [article]

Beren Millidge, Alexander Tschantz, Anil Seth, Christopher L Buckley
2020 arXiv   pre-print
for training deep networks.  ...  While motivated by high-level notions of variational inference, detailed neurophysiological models of cortical microcircuits which can implements its computations have been developed.  ...  BM would additionally like to thank Mycah Banks for her invaluable contribution in preparing the figures for this manuscript.  ... 
arXiv:2010.01047v2 fatcat:armb7kpybfgjbeoba2y3apdraa

A biologically plausible neural network for multi-channel Canonical Correlation Analysis [article]

David Lipshutz, Yanis Bahroun, Siavash Golkar, Anirvan M. Sengupta, Dmitri B. Chklovskii
2021 arXiv   pre-print
Cortical pyramidal neurons receive inputs from multiple distinct neural populations and integrate these inputs in separate dendritic compartments.  ...  We explore the possibility that cortical microcircuits implement Canonical Correlation Analysis (CCA), an unsupervised learning method that projects the inputs onto a common subspace so as to maximize  ...  is in contrast to cortical microcircuits where lateral influence between cortical pyramidal neurons is often indirect and mediated by interneurons.  ... 
arXiv:2010.00525v4 fatcat:ttpnnb7kfnablmldvfh3wxs4t4

A Biologically Plausible Neural Network for Multichannel Canonical Correlation Analysis

David Lipshutz, Yanis Bahroun, Siavash Golkar, Anirvan M. Sengupta, Dmitri B. Chklovskii
2021 Neural Computation  
Cortical pyramidal neurons receive inputs from multiple distinct neural populations and integrate these inputs in separate dendritic compartments.  ...  We explore the possibility that cortical microcircuits implement canonical correlation analysis (CCA), an unsupervised learning method that projects the inputs onto a common subspace so as to maximize  ...  is in contrast to cortical microcircuits where lateral influence between cortical pyramidal neurons is often indirect and mediated by interneurons.  ... 
doi:10.1162/neco_a_01414 pmid:34412114 fatcat:nqx6a4koovb6dg7nfq6cpzlttq

Anatomy and physiology of the thick-tufted layer 5 pyramidal neuron

Srikanth Ramaswamy, Henry Markram
2015 Frontiers in Cellular Neuroscience  
The thick-tufted layer 5 (TTL5) pyramidal neuron is one of the most extensively studied neuron types in the mammalian neocortex and has become a benchmark for understanding information processing in excitatory  ...  By virtue of having the widest local axonal and dendritic arborization, the TTL5 neuron encompasses various local neocortical neurons and thereby defines the dimensions of neocortical microcircuitry.  ...  Yun Wang for morphological reconstructions used in Figure 2 . This work was supported by the Blue Brain Project, EPFL, Switzerland.  ... 
doi:10.3389/fncel.2015.00233 pmid:26167146 pmcid:PMC4481152 fatcat:f5mmqcdfmjfflkqcqnjdczcbcu
« Previous Showing results 1 — 15 out of 122 results