Learning spatiotemporal signals using a recurrent spiking network that discretizes time

Maes, Amadeus; Barahona, Mauricio; Clopath, Claudia

doi:10.1371/journal.pcbi.1007606

Cited by 53 publications

(110 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Models for fast learning have been proposed, and in general they rely on a suited pre-training or preparation of the recurrent network [ 12 , 41 – 43 ]. This procedure is not task specific.…”

Section: Discussionmentioning

confidence: 99%

Target spike patterns enable efficient and biologically plausible learning for complex temporal tasks

2021

View full text Add to dashboard Cite

Recurrent spiking neural networks (RSNN) in the brain learn to perform a wide range of perceptual, cognitive and motor tasks very efficiently in terms of energy consumption and their training requires very few examples. This motivates the search for biologically inspired learning rules for RSNNs, aiming to improve our understanding of brain computation and the efficiency of artificial intelligence. Several spiking models and learning rules have been proposed, but it remains a challenge to design RSNNs whose learning relies on biologically plausible mechanisms and are capable of solving complex temporal tasks. In this paper, we derive a learning rule, local to the synapse, from a simple mathematical principle, the maximization of the likelihood for the network to solve a specific task. We propose a novel target-based learning scheme in which the learning rule derived from likelihood maximization is used to mimic a specific spatio-temporal spike pattern that encodes the solution to complex temporal tasks. This method makes the learning extremely rapid and precise, outperforming state of the art algorithms for RSNNs. While error-based approaches, (e.g. e-prop) trial after trial optimize the internal sequence of spikes in order to progressively minimize the MSE we assume that a signal randomly projected from an external origin (e.g. from other brain areas) directly defines the target sequence. This facilitates the learning procedure since the network is trained from the beginning to reproduce the desired internal sequence. We propose two versions of our learning rule: spike-dependent and voltage-dependent. We find that the latter provides remarkable benefits in terms of learning speed and robustness to noise. We demonstrate the capacity of our model to tackle several problems like learning multidimensional trajectories and solving the classical temporal XOR benchmark. Finally, we show that an online approximation of the gradient ascent, in addition to guaranteeing complete locality in time and space, allows learning after very few presentations of the target output. Our model can be applied to different types of biological neurons. The analytically derived plasticity learning rule is specific to each neuron model and can produce a theoretical prediction for experimental validation.

show abstract

Section: Discussionmentioning

confidence: 99%

Target spike patterns enable efficient and biologically plausible learning for complex temporal tasks

2021

View full text Add to dashboard Cite

show abstract

“…Interestingly, stability is encoded in time rather than space, which raises the question whether this approach could be combined with a pooling layer, reflecting temporal structure instead of spatial structure. Maes et al (2020) trained a recurrently connected spiking network such that small groups of neurons become active in succession and thus provide the basis for a simple index code. Via a supervisor signal, output neurons are trained to become responsive to a particular group or index from the recurrent network and, thus, fire in a temporal order encoded in the feed-forward weights to the output layer.…”

Section: Discussionmentioning

confidence: 99%

“…To improve stability, recent approaches used feed-forward structures (Pehlevan et al, 2018) or employed supervised learning rules (Laje and Buonomano, 2013). While feed-forward structures provide stable activity patterns, in general these play out on a very fast timescale (Zheng and Triesch, 2014) or require neural/synaptic adaptation such that activity moves between neuron groups (York and Van Rossum, 2009;Itskov et al, 2011;Murray et al, 2017;Maes et al, 2020). And since for supervised learning all states in the network need to be accessible at each computing unit, these so-called global learning rules are not compatible with most neuromorphic hardware.…”

Section: Introductionmentioning

confidence: 99%

Robust Trajectory Generation for Robotic Control on the Neuromorphic Research Chip Loihi

2020

View full text Add to dashboard Cite

Neuromorphic hardware has several promising advantages compared to von Neumann architectures and is highly interesting for robot control. However, despite the high speed and energy efficiency of neuromorphic computing, algorithms utilizing this hardware in control scenarios are still rare. One problem is the transition from fast spiking activity on the hardware, which acts on a timescale of a few milliseconds, to a control-relevant timescale on the order of hundreds of milliseconds. Another problem is the execution of complex trajectories, which requires spiking activity to contain sufficient variability, while at the same time, for reliable performance, network dynamics must be adequately robust against noise. In this study we exploit a recently developed biologically-inspired spiking neural network model, the so-called anisotropic network. We identified and transferred the core principles of the anisotropic network to neuromorphic hardware using Intel's neuromorphic research chip Loihi and validated the system on trajectories from a motor-control task performed by a robot arm. We developed a network architecture including the anisotropic network and a pooling layer which allows fast spike read-out from the chip and performs an inherent regularization. With this, we show that the anisotropic network on Loihi reliably encodes sequential patterns of neural activity, each representing a robotic action, and that the patterns allow the generation of multidimensional trajectories on control-relevant timescales. Taken together, our study presents a new algorithm that allows the generation of complex robotic movements as a building block for robotic control using state of the art neuromorphic hardware.

show abstract

“…Modelling studies so far have either focused on the study of sequential dynamics [38][39][40][41] or on motif acquisition [27][28][29]. This paper introduces an explicitly hierarchical model as a fundamental building block for the learning and replay of sequential dynamics of a compositional nature.…”

Section: From Serial To Hierarchical Modellingmentioning

confidence: 99%

“…Here, we present a model for learning temporal sequences on multiple scales implemented through a hierarchical network of bio-realistic spiking neurons and synapses. In contrast to current models, which focus on acquiring the motifs and speculate on the mechanisms to learn a syntax [27][28][29], our spiking network model learns motifs and syntax independently from a target sequence presented repeatedly. Furthermore, the plasticity of the synapses is entirely local, and does not rely on a global optimisation such as FORCE-training [30][31][32] or backpropagation through time [33].…”

Section: Introductionmentioning

confidence: 99%

Learning compositional sequences with multiple time scales through a hierarchical network of spiking neurons

2021

Self Cite

View full text Add to dashboard Cite

Sequential behaviour is often compositional and organised across multiple time scales: a set of individual elements developing on short time scales (motifs) are combined to form longer functional sequences (syntax). Such organisation leads to a natural hierarchy that can be used advantageously for learning, since the motifs and the syntax can be acquired independently. Despite mounting experimental evidence for hierarchical structures in neuroscience, models for temporal learning based on neuronal networks have mostly focused on serial methods. Here, we introduce a network model of spiking neurons with a hierarchical organisation aimed at sequence learning on multiple time scales. Using biophysically motivated neuron dynamics and local plasticity rules, the model can learn motifs and syntax independently. Furthermore, the model can relearn sequences efficiently and store multiple sequences. Compared to serial learning, the hierarchical model displays faster learning, more flexible relearning, increased capacity, and higher robustness to perturbations. The hierarchical model redistributes the variability: it achieves high motif fidelity at the cost of higher variability in the between-motif timings.

show abstract

Learning spatiotemporal signals using a recurrent spiking network that discretizes time

Cited by 53 publications

References 62 publications

Target spike patterns enable efficient and biologically plausible learning for complex temporal tasks

Target spike patterns enable efficient and biologically plausible learning for complex temporal tasks

Robust Trajectory Generation for Robotic Control on the Neuromorphic Research Chip Loihi

Learning compositional sequences with multiple time scales through a hierarchical network of spiking neurons

Contact Info

Product

Resources

About