Related Experiment Video
Updated: Jul 12, 2025

Photodiode-Based Optical Imaging for Recording Network Dynamics with Single-Neuron Resolution in Non-Transgenic Invertebrates
Published on: July 9, 2020
Event-driven adaptive optical neural network
Frank Brückerhoff-Plückelmann1, Ivonne Bente1, Marlon Becker2
1Physical Institute, University of Münster, Heisenbergstraße 11, 48149 Münster, Germany.
This study introduces an adaptive optical neural network with reconfigurable structures and non-volatile memory. The scalable photonic architecture demonstrates potential for advanced artificial intelligence hardware.
Area of Science:
- Photonics
- Artificial Intelligence
- Neuromorphic Computing
Background:
- Optical neural networks offer high-speed computation but often lack adaptability.
- Existing architectures struggle with dynamic reconfiguration and non-volatile memory integration.
Purpose of the Study:
- To develop a large-scale, adaptive optical neural network with structural plasticity.
- To implement non-volatile memory and nonlinear activation functions using phase-change materials.
Main Methods:
- Utilized wavelength-addressable artificial neurons with embedded phase-change materials.
- Employed multimode focusing for activation functions with excitatory and inhibitory responses.
- Trained the network using an evolutionary algorithm to classify text samples.
Main Results:
- Demonstrated a reversible switching contrast of 3.2 decibels for activation functions.
- Successfully trained a network of 736 subnetworks (8398 functional neurons) to distinguish between English and German text.
- Investigated both synaptic and structural plasticity during the training process.
Conclusions:
- The proposed adaptive optical neural network architecture is scalable and demonstrates significant potential for advanced AI applications.
- The integration of structural plasticity and non-volatile memory in photonic systems is feasible.
- Event-driven architectures combined with phase-change materials offer a promising path for next-generation neuromorphic computing.
More Related Videos
08:28Automated Multimodal Stimulation and Simultaneous Neuronal Recording from Multiple Small Organisms
Published on: March 3, 2023
07:12Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
Published on: April 11, 2025
Related Concept Videos
Neural Circuits
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
Vision
Photoreceptors and Visual Pathways
Neuroplasticity
Visual System
Once through the pupil, the light passes through the lens, a...
Light Acquisition