Emergence of associative learning in a neuromorphic inference network

Abstract

Objective. In the theoretical framework of predictive coding and active inference, the brain can be viewed as instantiating a rich generative model of the world that predicts incoming sensory data while continuously updating its parameters via minimization of prediction errors. While this theory has been successfully applied to cognitive processes – by modelling the activity of functional neural networks at a mesoscopic scale – the validity of the approach when modelling neurons as an ensemble of inferring agents, in a biologically plausible architecture, remained to be explored. Approach. We modelled a simplified cerebellar circuit with individual neurons acting as Bayesian agents to simulate the classical delayed eyeblink conditioning protocol. Neurons and synapses adjusted their activity to minimize their prediction error, which was used as the network cost function. This cerebellar network was then implemented in hardware by replicating digital neuronal elements via a low-power microcontroller. Main results. Persistent changes of synaptic strength – that mirrored neurophysiological observations – emerged via local (neurocentric) prediction error minimization, leading to the expression of associative learning. The same paradigm was effectively emulated in low-power hardware showing remarkably efficient performance compared to conventional neuromorphic architectures. Significance. These findings show that: i) an ensemble of free energy minimizing neurons – organized in a biological plausible architecture – can recapitulate functional self-organization observed in nature, such as associative plasticity, and ii) a neuromorphic network of inference units can learn unsupervised tasks without embedding predefined learning rules in the circuit, thus providing a potential avenue to a novel form of brain-inspired artificial intelligence.

Long-Term Synaptic Plasticity Tunes the Gain of Information Channels through the Cerebellum Granular Layer

“…The ability of LTP and LTD to reorganize granular layer activity bears a series of functional and theoretical consequences. LTP and LTD can finetune time- and frequency-dependent properties of network transmission rather than just regulating synaptic weights, bearing relevant implications for applicative scenarios [ 73 , 74 ]. Our observations lend support to the hypothesis that the granular layer processes spike bursts as a relevant computational unit, in which the composition in terms of the spike delay, the number or the frequency can be finetuned.…”

Emergence of brain-inspired small-world spiking neural network through neuroevolution

“…The replication of the cerebellum’s connectivity patterns using a low-power microcontroller, 51 confirms that the evolution of neural networks drives the tuning of synapses to reduce systemic free energy. From the perspective of energy consumption, the characteristics of the brain’s energy-saving operation mechanism under the immense pressure of information processing, including optimal input and noise, excitation/inhibition balance, the size of neurons and neuron clusters, have been analyzed.…”

Biologically plausible information propagation in a complementary metal-oxide semiconductor integrate-and-fire artificial neuron circuit with memristive synapses

“…This proof of principle is a first milestone in the development of advanced neuronal networks with performance compatible with brain circuits, given their capability to compute sparse and temporally uncorrelated information. Furthermore, differently from conventional hardware, neuromorphic electronic circuits [41] can be designed to operate with a limited power consumption in multiple time domains according to circuit architectures. These advantages, deriving from an electronic implementation of biologically plausible SNNs (e.g.…”