STDP in adaptive neurons gives close-to-optimal information transmission

Wulfram Gerstner, Guillaume Hennequin, Jean-Pascal Théodor Pfister
Frontiers Research Foundation, 2010
Article

Résumé

Spike-frequency adaptation is known to enhance the transmission of information in sensory spiking neurons by rescaling the dynamic range for input processing, matching it to the temporal statistics of the sensory stimulus. Achieving maximal information transmission has also been recently postulated as a role for spike-timing-dependent plasticity (STDP). However, the link between optimal plasticity and STDP in cortex remains loose, as does the relationship between STDP and adaptation processes. We investigate how STDP, as described by recent minimal models derived from experimental data, influences the quality of information transmission in an adapting neuron. We show that a phenomenological model based on triplets of spikes yields almost the same information rate as an optimal model specially designed to this end. In contrast, the standard pair-based model of STDP does not improve information transmission as much. This result holds not only for additive STDP with hard weight bounds, known to produce bimodal distributions of synaptic weights, but also for weight-dependent STDP in the context of unimodal but skewed weight distributions. We analyze the similarities between the triplet model and the optimal learning rule, and find that the triplet effect is an important feature of the optimal model when the neuron is adaptive. If STDP is optimized for information transmission, it must take into account the dynamical properties of the postsynaptic cell, which might explain the target-cell specificity of STDP. In particular, it accounts for the differences found in vitro between STDP at excitatory synapses onto principal cells and those onto fast-spiking interneurons.

Official source

https://infoscience.epfl.ch/record/161476?ln=fr

À propos de ce résultat

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Concepts associés

Chargement

Publications associées

Chargement

Wulfram Gerstner, Guillaume Hennequin, Jean-Pascal Théodor Pfister
Frontiers Research Foundation, 2010
Article

Résumé

Official source

https://infoscience.epfl.ch/record/161476?ln=fr

À propos de ce résultat

Concepts associés (18)

La (en Spike-timing-dependent plasticity, STDP) est un processus de modification du poids des synapses. Cette modification dépend du moment de déclenchement du potentiel d'action dans les neurones

An excitatory synapse is a synapse in which an action potential in a presynaptic neuron increases the probability of an action potential occurring in a postsynaptic cell. Neurons form networks throu

La transmission de données désigne le transport de données, quel que soit le type d'information, d'un endroit vers un autre, par un moyen physique ou numérique. Histoire Historiquement,

Afficher plus

Concepts associés

Chargement

Publications associées (52)

Publications associées

Chargement

Theory of representation learning in cortical neural networks

Carlos Stein Naves de Brito

Our brain continuously self-organizes to construct and maintain an internal representation of the world based on the information arriving through sensory stimuli. Remarkably, cortical areas related to different sensory modalities appear to share the same functional unit, the neuron, and develop through the same learning mechanism, synaptic plasticity. It motivates the conjecture of a unifying theory to explain cortical representational learning across sensory modalities. In this thesis we present theories and computational models of learning and optimization in neural networks, postulating functional properties of synaptic plasticity that support the apparent universal learning capacity of cortical networks. In the past decades, a variety of theories and models have been proposed to describe receptive field formation in sensory areas. They include normative models such as sparse coding, and bottom-up models such as spike-timing dependent plasticity. We bring together candidate explanations by demonstrating that in fact a single principle is sufficient to explain receptive field development. First, we show that many representative models of sensory development are in fact implementing variations of a common principle: nonlinear Hebbian learning. Second, we reveal that nonlinear Hebbian learning is sufficient for receptive field formation through sensory inputs. A surprising result is that our findings are independent of specific details, and allow for robust predictions of the learned receptive fields. Thus nonlinear Hebbian learning and natural statistics can account for many aspects of receptive field formation across models and sensory modalities. The Hebbian learning theory substantiates that synaptic plasticity can be interpreted as an optimization procedure, implementing stochastic gradient descent. In stochastic gradient descent inputs arrive sequentially, as in sensory streams. However, individual data samples have very little information about the correct learning signal, and it becomes a fundamental problem to know how many samples are required for reliable synaptic changes. Through estimation theory, we develop a novel adaptive learning rate model, that adapts the magnitude of synaptic changes based on the statistics of the learning signal, enabling an optimal use of data samples. Our model has a simple implementation and demonstrates improved learning speed, making this a promising candidate for large artificial neural network applications. The model also makes predictions on how cortical plasticity may modulate synaptic plasticity for optimal learning. The optimal sampling size for reliable learning allows us to estimate optimal learning times for a given model. We apply this theory to derive analytical bounds on times for the optimization of synaptic connections. First, we show this optimization problem to have exponentially many saddle-nodes, which lead to small gradients and slow learning. Second, we show that the number of input synapses to a neuron modulates the magnitude of the initial gradient, determining the duration of learning. Our final result reveals that the learning duration increases supra-linearly with the number of synapses, suggesting an effective limit on synaptic connections and receptive field sizes in developing neural networks.

EPFL2016

Chargement

Generalized Bienenstock-Cooper-Munro rule for spiking neurons that maximizes information transmission

Wulfram Gerstner, Jean-Pascal Théodor Pfister, Taro Toyoizumi

Maximization of information transmission by a spiking-neuron model predicts changes of synaptic connections that depend on timing of pre- and postsynaptic spikes and on the postsynaptic membrane potential. Under the assumption of Poisson firing statistics, the synaptic update rule exhibits all of the features of the Bienenstock-Cooper-Munro rule, in particular, regimes of synaptic potentiation and depression separated by a sliding threshold. Moreover, the learning rule is also applicable to the more realistic case of neuron models with refractoriness, and is sensitive to correlations between input spikes, even in the absence of presynaptic rate modulation. The learning rule is found by maximizing the mutual information between presynaptic and postsynaptic spike trains under the constraint that the postsynaptic firing rate stays close to some target firing rate. An interpretation of the synaptic update rule in terms of homeostatic synaptic processes and spike-timing-dependent plasticity is discussed.

2005

Chargement

In the present work, we approach two key aspects of memory formation: associative memory and synaptic consolidation. The storage of associative memory is commonly related to the medial temporal lobe in humans. Experimental evidence shows that the memories of objects, people or places are represented in this brain area by cell assemblies that respond selectively to single concepts. Neurons forming such assemblies are called concept cells. Associations between different concepts are linked to concept cells shared between assemblies: we refer to the number of shared neurons as the overlap between memory engrams. The respective assemblies of two associated concepts (e.g. Hillary and Bill Clinton) share more neurons compared to the assemblies of two unrelated concepts (e.g. Hillary Clinton and the Eiffel tower). In particular, three characteristics of assemblies of concept cells are important for this work: (a) they exhibit a very low mean activity (about 0.2% of neurons respond to each concept), (b) overlapping assemblies share about 4% of their cells, (c) non-overlapping assemblies share less then 1% of their cells. This implies that the association between two concepts induces a higher level of overlap between the relative memory engrams. In parallel, theoretical studies have shown that overlaps between memory engrams are fundamental in the process of free recall of sequences of words. These models assume that memory engrams have such a high mean activity that all assemblies are overlapping to a certain extent. Associative memory is traditionally modeled through attractor neural networks. Memory engrams are represented by binary patterns of active/silent neurons. While there is extensive literature on independent low-activity patterns, only a few studies can be found on correlated patterns. Extending the existing theory to include correlation is a key missing point to answer questions such as: How do shared neurons encode association? Why are 4% of neurons shared and not more? Using a mean-field approximation, we derive analytic equations for the network dynamics in the case of correlated patterns. Our results provide a theoretical framework that can explain the experimentally observed value of shared neurons. We find that for concepts represented by realistically sparse neural assemblies there are a minimal and a maximal fraction of shared neurons so that associations can be reliably coded. In the presence of a periodically modulated signal, such as hippocampal oscillations, chains of associations can be recalled analogously to theories of free recall of lists of memorized words. Finally, we compare the predicted number of concepts a neuron responds to with experimental data. We test different ways of constructing correlated patterns and confirmed the common opinion that information in the hippocampus is non-hierarchically organised. In the second part of the thesis, we propose a model of synaptic consolidation based on two coupled dynamical variables: the fast synaptic weights and a slow internal synaptic mechanism. In classical experiments, the consolidation of the synapse is related to the stimulation frequency and number of repetitions. We show that it is exactly the time scale separation between the dynamics of the two variables that determines which combination of stimulation amplitude and frequency are suitable to elicit synaptic consolidation.

EPFL2021

Chargement

Theory of representation learning in cortical neural networks

Carlos Stein Naves de Brito

EPFL2016

EPFL2021