Spiking neuron bridges the gap between biological and artificial neurons

Engineers from the University of Liège have made significant advances in the design of artificial neural networks by developing a new form of spiking neuron. The Spiking Recurrent Cell (SRC) is a novel model that combines ease of implementation with the capacity to replicate the dynamics of biological neurons.

This novel model provides intriguing new opportunities for neuro-inspired artificial intelligence, on top of spiking neurons' energy efficiency. The study has been published as an open-access study in Neuromorphic Computing and Engineering (IOP Science).

Two forms of neural networks are utilized in artificial intelligence: artificial neural networks (ANNs) and spiking neural networks (SNNs). Their uses, functions, and structures, however, vary considerably. ANNs are comparatively easier to develop and are used widely for a range of machine learning applications (games, speech recognition, image recognition, etc.). Nevertheless, they are computationally costly and inefficient in terms of energy.

SNNs, on the other hand, provide more accurate modeling of biological neural processes and are employed in applications that need sensitivity to the exact timing of events (such as robotics, brain-computer interface, and sensory processing). They are not like artificial neural networks (ANNs) in that they imitate the functioning of biological neurons by using impulses, or spikes, as the only means of communication between neurons.

When these SNNs are run on specific hardware - known as neuromorphic hardware - their energy consumption becomes extremely low. This characteristic means that such networks can be used in situations where energy efficiency is paramount, such as in embedded systems, which are autonomous computer and electronic systems that perform a precise task within the device in which they are integrated.”

Florent De Geeter, Research Engineer, Montefiore Institute, University of Liège

Since SNNs are harder to train than ANNs, current research focuses on creating training methods that will allow SNNs to compete with ANNs on challenging tasks.

A novel approach has been attempted as part of an ambitious project at ULiège: by altering the dynamics of a well-known class of artificial neurons that is easy to train, researchers have succeeded in simulating the behavior of biological neurons, leading to the creation of a new model: the Spiking Recurrent Cell (SRC).

SRC: A Bridge between ANNs and SNNs

The major innovation of this study lies in the design of this Spiking Recurrent Cell (SRC) a neuron model capable of generating spikes autonomously, like biological neurons. Unlike conventional SNN models where spikes are generated artificially, the SRC model allows for a more natural and dynamic emulation of neuronal impulses.”

Damien Ernst, Study Co-Author and Professor, University of Liège

The complex learning algorithms of ANNs and the energy efficiency of SNNs can now be combined, thanks to this new paradigm. By integrating the benefits of both types of neural networks, SRCs provide a hybrid solution and open the door for the development of new SNNs.

Implications and Future Applications

SRCs have a wide range of possible uses. SNNs can be applied in situations where energy consumption is crucial, such the on-board systems of autonomous vehicles, because of their energy efficiency.

Furthermore, the ability of the SRC model to simulate various neuronal behaviors by adjusting its internal parameters makes these networks more expressive and closer to biological networks, enabling significant advances in the understanding and reproduction of brain functions.”

Guillaume Drion, Study Co-Author and Director, Neuromorphic Engineering Laboratory, University of Liège

The development and introduction of the Spiking Recurrent Cell marks a big step forward in neural network research, combining the capabilities of ANNs and SNNs. This breakthrough presents new avenues for the creation of more efficient and energy-saving intelligent systems.

Journal reference:

‌De Geeter, F., et al. (2024) Spike-based computation using classical recurrent neural networks. Neuromorphic Computing and Engineering. doi.org/10.1088/2634-4386/ad473b


The opinions expressed here are the views of the writer and do not necessarily reflect the views and opinions of AZoLifeSciences.
Post a new comment

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.

You might also like...
New ProteoCHIP EVO 96 Boosts Single-Cell Proteomics Sensitivity and Throughput