Source author record

Wulfram Gerstner

Wulfram Gerstner appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Neurons and Cognition Machine Learning Neural and Evolutionary Computing eess.AS Sound

Catalog footprint

What is connected

12works

5topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Self-supervised local learning rules learn the hidden hierarchical structure of high-dimensional data

The brain learns abstract representations of high-dimensional sensory input, but the plasticity rules that enable such learning are unknown. We study biologically plausible algorithms on the Random Hierarchy Model (RHM), an artificial dataset designed to investigate how deep neural networks learn the intrinsic hierarchical structure of high-dimensional data. We focus on two types of local learning rules that avoid both a long convergence time and the use of a symmetric error network. The first type uses direct feedback signals to approximate error propagation from the output layer. The second type uses layerwise self-supervised contrastive or non-contrastive loss functions that do not explicitly approximate errors at the output layer. We show that all rules of the first type fail to solve the tasks of the RHM and trace this failure back to input-specific nonlinearities (`masking') that are implemented in full backpropagation and are essential for learning complex tasks. However, algorithms of the second type are able to learn the hierarchical hidden structure of the RHM tasks and are as data-efficient as supervised backpropagation training, while being compatible with known rules of synaptic plasticity in cortex.

preprint2023arXiv

Mesoscopic modeling of hidden spiking neurons

Can we use spiking neural networks (SNN) as generative models of multi-neuronal recordings, while taking into account that most neurons are unobserved? Modeling the unobserved neurons with large pools of hidden spiking neurons leads to severely underconstrained problems that are hard to tackle with maximum likelihood estimation. In this work, we use coarse-graining and mean-field approximations to derive a bottom-up, neuronally-grounded latent variable model (neuLVM), where the activity of the unobserved neurons is reduced to a low-dimensional mesoscopic description. In contrast to previous latent variable models, neuLVM can be explicitly mapped to a recurrent, multi-population SNN, giving it a transparent biological interpretation. We show, on synthetic spike trains, that a few observed neurons are sufficient for neuLVM to perform efficient model inversion of large SNNs, in the sense that it can recover connectivity parameters, infer single-trial latent population activity, reproduce ongoing metastable dynamics, and generalize when subjected to perturbations mimicking photo-stimulation.

preprint2022arXiv

Correlation-invariant synaptic plasticity

Cortical populations of neurons develop sparse representations adapted to the statistics of the environment. While existing synaptic plasticity models reproduce some of the observed receptive-field properties, a major obstacle is the sensitivity of Hebbian learning to omnipresent spurious correlations in cortical networks which can overshadow relevant latent input features. Here we develop a theory for synaptic plasticity that is invariant to second-order correlations in the input. Going beyond classical Hebbian learning, we show how Hebbian long-term depression (LTD) cancels the sensitivity to second-order correlations, so that receptive fields become aligned with features hidden in higher-order statistics. Our simulations demonstrate how correlation-invariance enables biologically realistic models to develop sparse population codes, despite diverse levels of variability and heterogeneity. The theory advances our understanding of local unsupervised learning in cortical circuits and assigns a specific functional role to synaptic LTD mechanisms in pyramidal neurons.

preprint2019arXiv

Learning to Generate Music with BachProp

As deep learning advances, algorithms of music composition increase in performance. However, most of the successful models are designed for specific musical structures. Here, we present BachProp, an algorithmic composer that can generate music scores in many styles given sufficient training data. To adapt BachProp to a broad range of musical styles, we propose a novel representation of music and train a deep network to predict the note transition probabilities of a given music corpus. In this paper, new music scores generated by BachProp are compared with the original corpora as well as with different network architectures and other related models. We show that BachProp captures important features of the original datasets better than other models and invite the reader to a qualitative comparison on a large collection of generated songs.

preprint2016arXiv

Algorithmic Composition of Melodies with Deep Recurrent Neural Networks

A big challenge in algorithmic composition is to devise a model that is both easily trainable and able to reproduce the long-range temporal dependencies typical of music. Here we investigate how artificial neural networks can be trained on a large corpus of melodies and turned into automated music composers able to generate new melodies coherent with the style they have been trained on. We employ gated recurrent unit networks that have been shown to be particularly efficient in learning complex sequential activations with arbitrary long time lags. Our model processes rhythm and melody in parallel while modeling the relation between these two features. Using such an approach, we were able to generate interesting complete melodies or suggest possible continuations of a melody fragment that is coherent with the characteristics of the fragment itself.

preprint2016arXiv

Towards deep learning with spiking neurons in energy based models with contrastive Hebbian plasticity

In machine learning, error back-propagation in multi-layer neural networks (deep learning) has been impressively successful in supervised and reinforcement learning tasks. As a model for learning in the brain, however, deep learning has long been regarded as implausible, since it relies in its basic form on a non-local plasticity rule. To overcome this problem, energy-based models with local contrastive Hebbian learning were proposed and tested on a classification task with networks of rate neurons. We extended this work by implementing and testing such a model with networks of leaky integrate-and-fire neurons. Preliminary results indicate that it is possible to learn a non-linear regression task with hidden layers, spiking neurons and a local synaptic plasticity rule.

preprint2015arXiv

Fluctuations and information filtering in coupled populations of spiking neurons with adaptation

Finite-sized populations of spiking elements are fundamental to brain function, but also used in many areas of physics. Here we present a theory of the dynamics of finite-sized populations of spiking units, based on a quasi-renewal description of neurons with adaptation. We derive an integral equation with colored noise that governs the stochastic dynamics of the population activity in response to time-dependent stimulation and calculate the spectral density in the asynchronous state. We show that systems of coupled populations with adaptation can generate a frequency band in which sensory information is preferentially encoded. The theory is applicable to fully as well as randomly connected networks, and to leaky integrate-and-fire as well as to generalized spiking neurons with adaptation on multiple time scales.

preprint2013arXiv

Reward-based learning under hardware constraints - Using a RISC processor embedded in a neuromorphic substrate

In this study, we propose and analyze in simulations a new, highly flexible method of implementing synaptic plasticity in a wafer-scale, accelerated neuromorphic hardware system. The study focuses on globally modulated STDP, as a special use-case of this method. Flexibility is achieved by embedding a general-purpose processor dedicated to plasticity into the wafer. To evaluate the suitability of the proposed system, we use a reward modulated STDP rule in a spike train learning task. A single layer of neurons is trained to fire at specific points in time with only the reward as feedback. This model is simulated to measure its performance, i.e. the increase in received reward after learning. Using this performance as baseline, we then simulate the model with various constraints imposed by the proposed implementation and compare the performance. The simulated constraints include discretized synaptic weights, a restricted interface between analog synapses and embedded processor, and mismatch of analog circuits. We find that probabilistic updates can increase the performance of low-resolution weights, a simple interface between analog synapses and processor is sufficient for learning, and performance is insensitive to mismatch. Further, we consider communication latency between wafer and the conventional control computer system that is simulating the environment. This latency increases the delay, with which the reward is sent to the embedded processor. Because of the time continuous operation of the analog synapses, delay can cause a deviation of the updates as compared to the not delayed situation. We find that for highly accelerated systems latency has to be kept to a minimum. This study demonstrates the suitability of the proposed implementation to emulate the selected reward modulated STDP learning rule.

preprint2013arXiv

Spike timing prediction with active dendrites

A complete single-neuron model must correctly reproduce the firing of spikes and bursts. We present a study of a simplified model of deep pyramidal cells of the cortex with active dendrites. We hypothesized that we can model the soma and its apical tuft with only two compartments, without significant loss in the accuracy of spike-timing predictions. The model is based on experimentally measurable impulse-response functions, which transfer the effect of current injected in one compartment to current reaching the other. Each compartment was modeled with a pair of non-linear differential equations and a small number of parameters that approximate the Hodgkin-and-Huxley equations. The predictive power of this model was tested on electrophysiological experiments where noisy current was injected in both the soma and the apical dendrite simultaneously. We conclude that a simple two-compartment model can predict spike times of pyramidal cells stimulated in the soma and dendrites simultaneously. Our results support that regenerating activity in the dendritic tuft is required to properly account for the dynamics of layer 5 pyramidal cells under in-vivo-like conditions.

preprint2012arXiv

Nonnormal amplification in random balanced neuronal networks

In dynamical models of cortical networks, the recurrent connectivity can amplify the input given to the network in two distinct ways. One is induced by the presence of near-critical eigenvalues in the connectivity matrix W, producing large but slow activity fluctuations along the corresponding eigenvectors (dynamical slowing). The other relies on W being nonnormal, which allows the network activity to make large but fast excursions along specific directions. Here we investigate the tradeoff between nonnormal amplification and dynamical slowing in the spontaneous activity of large random neuronal networks composed of excitatory and inhibitory neurons. We use a Schur decomposition of W to separate the two amplification mechanisms. Assuming linear stochastic dynamics, we derive an exact expression for the expected amount of purely nonnormal amplification. We find that amplification is very limited if dynamical slowing must be kept weak. We conclude that, to achieve strong transient amplification with little slowing, the connectivity must be structured. We show that unidirectional connections between neurons of the same type together with reciprocal connections between neurons of different types, allow for amplification already in the fast dynamical regime. Finally, our results also shed light on the differences between balanced networks in which inhibition exactly cancels excitation, and those where inhibition dominates.

preprint2010arXiv

Rescaling, thinning or complementing? On goodness-of-fit procedures for point process models and Generalized Linear Models

Generalized Linear Models (GLMs) are an increasingly popular framework for modeling neural spike trains. They have been linked to the theory of stochastic point processes and researchers have used this relation to assess goodness-of-fit using methods from point-process theory, e.g. the time-rescaling theorem. However, high neural firing rates or coarse discretization lead to a breakdown of the assumptions necessary for this connection. Here, we show how goodness-of-fit tests from point-process theory can still be applied to GLMs by constructing equivalent surrogate point processes out of time-series observations. Furthermore, two additional tests based on thinning and complementing point processes are introduced. They augment the instruments available for checking model adequacy of point processes as well as discretized models.

preprint2004arXiv

Predicting spike times of a detailed conductance-based neuron model driven by stochastic spike arrival

Reduced models of neuronal activity such as Integrate-and-Fire models allow a description of neuronal dynamics in simple, intuitive terms and are easy to simulate numerically. We present a method to fit an Integrate-and-Fire-type model of neuronal activity, namely a modified version of the Spike Response Model, to a detailed Hodgkin-Huxley-type neuron model driven by stochastic spike arrival. In the Hogkin-Huxley model, spike arrival at the synapse is modeled by a change of synaptic conductance. For such conductance spike input, more than 70% of the postsynaptic action potentials can be predicted with the correct timing by the Integrate-and-Fire-type model. The modified Spike Response Model is based upon a linearized theory of conductance-driven Integrate-and-Fire neuron.