Simplicity and efficiency of integrate-and-fire neuron models

Area of Science:

• Computational neuroscience
• Neural network modeling

Background:

• The very simple spiking neuron (VSSN) model was proposed as an efficient alternative to the integrate-and-fire (I&F) neuron model for large neural network simulations.
• Recent advances in neuronal network modeling have introduced techniques for efficient state computation and exact subthreshold dynamics integration.

Purpose of the Study:

• To evaluate the efficiency and suitability of the VSSN model compared to state-of-the-art methods in neuronal network modeling.
• To highlight the limitations of the VSSN model in light of recent computational neuroscience advancements.

Main Methods:

• Comparative analysis of the VSSN model against advanced integrate-and-fire neuron model solvers.
• Evaluation of computational efficiency in simulating large neuronal networks.

Main Results:

• The VSSN model does not incorporate key advances in neuronal network modeling, such as efficient state computation and exact subthreshold dynamics integration.
• State-of-the-art solvers for I&F neuron models demonstrate substantially greater efficiency than the VSSN simulator.

Conclusions:

• The VSSN model is not an efficient replacement for the integrate-and-fire neuron model for large-scale simulations.
• Current advanced solvers enable routine simulations of large networks (10^5 neurons, 10^9 connections) on moderate computing clusters, surpassing VSSN's capabilities.