Simulation and Modeling Methodologies: Enabler for Neuromorphic Computing Applications

2021 
Neuromorphic computing is of worldwide interest. Compared to the von Neumann’s computer architecture, neuromorphic systems offer advantages and novel approaches for artificial intelligence problems to be solved. Inspired by biology, neuromorphic systems adopt the theory of the human brain modeling by implementing neurons and synapses with the help electronic devices and circuits. Many researchers developed new algorithms, learning approaches, models, etc., implement them into hardware to explore the neuromorphic system. However, many of the promising approaches concentrate on the realization not taking into account the feasibility for industrial or consumer application of the various concepts.Here, simulation and modeling methodologies are discussed with a bench of examples of different applications from well know domains, e.g. MEMS, IC, etc. An overview is given where and when the different approaches/methodologies makes sense, starting from scratch towards predictive simulations for detailed analysis and the needs for realization in mass production. Afterwards, discussion is continued towards neuromorphic computing systems. In this paper we would like to draw the attention of the reader why it makes sense to use the support of such methods and why it is so important to push the development of simulation and modeling for neuromorphic computing systems.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    42
    References
    0
    Citations
    NaN
    KQI
    []