Learning spatiotemporal signals using a recurrent spiking network that discretizes time

Abstract

Learning to produce spatiotemporal sequences is a common task the brain has to solve. The same neural substrate may be used by the brain to produce different sequential behaviours. The way the brain learns and encodes such tasks remains unknown as current computational models do not typically use realistic biologicallyplausible learning. Here, we propose a model where a spiking recurrent network of excitatory and inhibitory biophysical neurons drives a read-out layer: the dynamics of the recurrent network is constrained to encode time while the read-out neurons encode space. Space is then linked with time through plastic synapses that follow common Hebbian learning rules. We demonstrate that the model is able to learn spatiotemporal dynamics on a timescale that is behaviourally relevant. Learned sequences are robustly replayed during a regime of spontaneous activity. Author summaryThe brain has the ability to learn flexible behaviours on a wide range of time scales. Previous studies have successfully build spiking network models that learn a variety of computational tasks. However, often the learning involved is not local. Here, we investigate a model using biological-plausible plasticity rules for a specific computational task: spatiotemporal sequence learning. The architecture separates time and space into two different parts and this allows learning to bind space to time. Importantly, the time component is encoded into a recurrent network which exhibits sequential dynamics on a behavioural time scale. This network is then used as an engine to drive spatial read-out neurons. We demonstrate that the model can learn complicated spatiotemporal spiking dynamics, such as the song of a bird, and replay the song robustly.

Frenesy: Time-symmetric dynamical activity in nonequilibria

“…By relying on gradient flow in free energy landscapes, the largest part of neuro-computational models do not typically use realistic biologically-plausible processes, while time "in our brain" is probably again nothing but movement; see e.g. [136] for a recent and workable encoding of time in a neural network.…”

Learning precise spatiotemporal sequences via biophysically realistic circuits with modular structure

“…proposed, based on a periodic and essentially deterministic timing network, which also uses biologically plausible learning rules to learn to represent single non-Markovian sequences 45 .…”

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

“…The design of the clock networks follows Ref. Maes et al 2020. Each clock is composed of clusters of excitatory neurons coupled in a cycle with a directional bias (i.e., neurons in cluster are more strongly connected to neurons in cluster + 1) together with a central cluster of inhibitory neurons coupled to all the excitatory clusters ( Fig.…”

“…The weights in the fast and slow clocks and the interneuron wiring are assumed to originate from earlier processes during evolution or early development. Previous computational studies have shown that sequential dynamics can be learnt in recurrent networks, both in an unsupervised (Jun and Jin 2007; Zheng and Triesch 2014) and supervised (Murray and Escola 2017;Maes et al 2020) fashion.…”

“…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 (Stroud et al 2018;Logiaco et al 2019;Maes et al 2020), 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 (Nicola and Clopath 2017;Hardy and Buonomano 2018;Nicola and Clopath 2019) or backpropagation through time (Werbos 1990).…”