Type
[tba]
Prerequisites
[tba, please inquire]
Description
Spiking Neural Networks (SNNs) are a more biologically realistic approach to artificial neural networks that closely mimic how neurons communicate in biological brains. Unlike traditional artificial neural networks that transmit continuous values, SNNs use discrete spikes or pulses to transmit information, similar to actual neurons' action potentials. These networks incorporate the concept of time explicitly, as neurons accumulate charge over time and only fire when reaching a certain threshold. This temporal aspect allows SNNs to process information more efficiently in terms of energy consumption and potentially enables more sophisticated temporal pattern recognition. While SNNs show promise for applications in neuromorphic computing and brain-machine interfaces, they currently face challenges in training and implementation due to their complex dynamics and the discrete nature of spike signals. Despite these challenges, they represent an important bridge between neuroscience and artificial intelligence, offering insights into both biological neural processing and new computing paradigms.
References
- Spiking neuron models as biologically inspired computational models
Spiking neuron models: single neurons, populations, plasticity https://lcnwww.epfl.ch/gerstner/PUBLICATIONS/SpikingNeuronM-extracts.pdf
Networks of spiking neurons: The third generation of neural network models https://igi-web.tugraz.at/PDF/85a.pdf - A survey on the training spiking neural networks
Training Spiking Neural Networks Using Lessons From Deep Learning https://arxiv.org/abs/2109.12894 - Expressivity
Expressivity of Spiking Neural Networks https://arxiv.org/abs/2308.08218 - Energy efficiency
To Spike or Not To Spike: A Digital Hardware Perspective on Deep Learning Acceleration https://arxiv.org/pdf/2306.15749 - Neuromorphic computing
Memory and information processing in neuromorphic systems.https://arxiv.org/pdf/1506.03264 - Applications: SNNs for Vision
Unsupervised learning of digit recognition using spike-timing-dependent plasticity. https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2015.00099/full