Related Experiment Video
Updated: Jul 16, 2025
10:50
Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches
Published on: June 21, 2022
1.8K
Sensitivity analysis of point neuron model simulations implemented on neuromorphic hardware
Srijanie Dey1, Alexander G Dimitrov1
1Department of Mathematics, Washington State University, Vancouver, WA, United States.
Frontiers in Neuroscience
|September 11, 2023
Summary
Intel
Area of Science:
- Neuro-inspired computing
- Computational neuroscience
Background:
- Neuromorphic computing architectures require rigorous validation.
- Comparing neuromorphic platforms with conventional simulations is crucial.
Purpose of the Study:
- To establish a framework for comparing neuromorphic and conventional computing platforms.
- To assess the robustness of neuromorphic models via sensitivity analysis.
Main Methods:
- Utilized the Leaky Integrate and Fire (LIF) model with mouse visual cortex data.
- Validated Intel's Loihi neuromorphic chip against classical simulations.
- Performed sensitivity analysis by varying model parameters individually and in pairs.
Main Results:
- Loihi demonstrated efficient and precise replication of classical simulations.
- The neuromorphic model showed robustness across most parameter variations.
- Identified specific parameters requiring precise definition for model sensitivity.
Conclusions:
- The developed methodology enables effective comparison of neuromorphic and conventional platforms.
- Intel's Loihi chip shows high fidelity in simulating neuronal dynamics.
- Sensitivity analysis highlights critical parameters for future neuromorphic model refinement.
