Wavelength-Specific Diffractive Neural Networks for Crosstalk Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optical diffractive deep neural networks experience crosstalk due to light of different wavelength regions, leading to reduced accuracy in target object detection.
Innovation Solution
The neural network device includes multiple optical diffractive deep neural networks optimized for specific wavelength regions, a light guide unit to direct light to these networks, a light receiving portion to capture the output, and a control unit to detect objects based on the received light signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If optical diffractive deep neural network processes light of different wavelength regions, then the detection capability is enhanced, but crosstalk occurs and detection accuracy deteriorates
Solution Approach 1:
The optical diffractive deep neural network is divided into multiple wavelength-specific networks, each optimized for a particular wavelength region. This segmentation prevents crosstalk between different wavelength regions while maintaining the ability to detect multiple types of target objects across different spectral bands.
Solution Approach 2:
Each optical diffractive deep neural network is optimized with local quality specific to its designated wavelength region, including tailored heterogeneous phase distributions and structural parameters. This ensures maximum detection accuracy for each wavelength band without interference from other wavelengths.
2Reliability
If optical diffractive deep neural network uses heterogeneous phases for light processing, then detection functionality is improved, but crosstalk between wavelength regions increases
Solution Approach 1:
The system segments the heterogeneous phases into separate optical diffractive deep neural networks, each containing heterogeneous phases optimized for a specific wavelength region. This segmentation isolates the harmful crosstalk effect while preserving the beneficial detection functionality of heterogeneous phase interactions.
Solution Approach 2:
The patent introduces wavelength-selective optical elements as intermediaries between the light source and the optical diffractive deep neural networks. These intermediaries filter and direct specific wavelength regions to appropriate networks, preventing crosstalk while maintaining comprehensive detection capability.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This configuration reduces crosstalk, enhances accuracy, and enables high-speed, energy-efficient target object detection by optimizing light processing for specific wavelength regions.
Implementation Method 1
the input light is repeatedly reflected, diffracted, and absorbed by heterogeneous phases mixed inside before being emitted from the emission surface
Implementation Method 2
the input light is repeatedly reflected, diffracted, and absorbed by heterogeneous phases mixed inside
Implementation Method 3
a light guide unit that guides light of an optimized wavelength to the optical diffractive deep neural network
Implementation Method 4
a light receiving portion that receives light output from the optical diffractive deep neural network
Data Source
AI summary
A neural network device includes one or a plurality of optical diffractive deep neural networks each optimized for light of a predetermined wavelength region, a light guide unit that guides light of an optimized wavelength region to the optical diffractive deep neural network, a light receiving portion that receives light output from the optical diffractive deep neural network, and a control unit that detects a target object on the basis of a signal corresponding to the light received by the light receiving portion.