Biomimetic model of the outer plexiform layer by incorporating memristive devices

Area of Science:

• Neuromorphic engineering
• Computational neuroscience
• Materials science

Background:

• The outer plexiform layer (OPL) in vertebrate retinas performs crucial early visual processing.
• Memristive nanodevices offer unique nonlinear and adaptive properties suitable for bio-inspired computing.

Purpose of the Study:

• To develop a biorealistic neuromorphic model for early vision processing using memristive nanodevices.
• To emulate the functional organization and dynamics of the vertebrate OPL.
• To evaluate the performance of memristor-based systems for image processing tasks.

Main Methods:

• Proposed a neuromorphic network architecture based on the OPL.
• Utilized hexagonal memristive grids to emulate OPL smoothing.
• Implemented a memristor-based thresholding scheme for edge detection.
• Assessed system adaptation and fault tolerance under varying conditions.

Main Results:

• Demonstrated memristive devices as effective building blocks for ultradense neuromorphic networks.
• Showcased hexagonal memristive grids for enhancing dynamic range via OPL smoothing emulation.
• Successfully employed memristor-based thresholding for grayscale image edge detection.
• Evaluated the system's robustness to light/noise variations and device yield.

Conclusions:

• Memristive nanodevices are valuable for creating neuromorphic architectures with biological-like dynamics.
• The proposed model effectively emulates OPL functions for enhanced image processing.
• The memristor-based system exhibits promising adaptation and fault tolerance capabilities.