Topic: spiking networks

Gaute Einevoll and Mihai Petrovici discuss how neuromorphic hardware can support artificial intelligence and efficient brain simulation. They contrast high-frequency arithmetic on conventional silicon with the rich dynamics and millisecond spikes of biological neurons, and explore brain-inspired systems for emulating neural dynamics and studying computation in spiking networks. Petrovici is a senior researcher at the University of Bern. The conversation was recorded on 18 June 2026 and released on 23 September 2026; the publisher’s audio runs 1 hour, 41 minutes and 46 seconds.

Friedemann Zenke discusses learning in spiking networks and the challenges of translating machine-learning methods into biologically plausible rules.

The spike-threshold nonlinearity is a fundamental, yet enigmatic, component of biological computation — despite its role in many theories, it has evaded definitive characterisation. Indeed, much classic work has attempted to limit the focus on spiking by smoothing over the spike threshold or by approximating spiking dynamics with firing-rate dynamics. Here, we take a novel perspective that captures the full potential of spike-based computation. Based on previous studies of the geometry of efficient spike-coding networks, we consider a population of neurons with low-rank connectivity, allowing

Many engineered and biological systems must trade off performance and energy use, and the brain is no exception. While there are theories on how activity levels are controlled in biological networks through feedback control (homeostasis), it is not clear what the effects on population coding are, and therefore how performance and energy can be traded off. In this talk we will consider this tradeoff in auto-encoding networks, in which there is a clear definition of performance (the coding loss). We first show how SNNs follow a characteristic trade-off curve between activity levels and coding lo

Summary: Modulation of neural activity by hormones, peptides, and other small molecules is known to be required for learning, motor control, and homeostasis, and it can induce similarly structured networks to exhibit different behaviors. Yet, relatively few theoretic studies have attempted to detangle the role and function the various types of modulation play throughout the brain. In particular, it is important to understand how changing neuromodulator concentrations can move a network through its space of possible behaviors. A better understanding of neuromodulatory movement through behaviora

Neural circuits exhibit complex activity patterns, both spontaneously and in response to external stimuli. Information encoding and learning in neural circuits depend on the ability of time-varying stimuli to control spontaneous network activity. In particular, variability arising from the sensitivity to initial conditions of recurrent cortical circuits can limit the information conveyed about the sensory input. Spiking and firing rate network models can exhibit such sensitivity to initial conditions that are reflected in their dynamic entropy rate and attractor dimensionality computed from th

Neural systems are remarkably robust against various perturbations, a phenomenon that still requires a clear explanation. Here, we graphically illustrate how neural networks can become robust. We study spiking networks that generate low-dimensional representations, and we show that the neurons’ subthreshold voltages are confined to a convex region in a lower-dimensional voltage subspace, which we call a ‘bounding box.’ Any changes in network parameters (such as number of neurons, dimensionality of inputs, firing thresholds, synaptic weights, or transmission delays) can all be understood as def

Since the popularization of GPUs for machine learning (ML) workloads, several dedicated accelerator chips have emerged, offering architectures optimized for common ML operations and requirements. The types of tasks targeted by such platforms however - often sparse and highly parallel - are not confined to the realm of traditional layer-based learning. Established simulators for large-scale spiking networks with biologically plausible connectivity and synaptic density have historically targeted CPUs, with GPU support and ports being a relatively recent development[1][2]. Here, we present a work

Coherent oscillations of neural activity are ubiquitous across brain spatial and temporal scales and have been associated with the formation of sensory or behavioral representations [1]. Refined experimental techniques report today a new complexity of this oscillatory activity, especially in the gamma (30-100 Hz) band. Indeed, the frequency of these oscillations shows a large variability through time [2], questioning the classical view of γ oscillations being separated into different frequency bands. Oscillations appear instead in bursts characterised by rapid, synchronous neuronal firing. Thi