Neural Network Circuit Using Electromagnetic Radiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scalability and complexity issues arise in neural networks due to the need for numerous electrical connections between nodes, particularly as the number of nodes increases, making hardware implementation challenging.
Innovation Solution
A circuit arrangement utilizing a common passband for electromagnetic radiation transmission between nodes, eliminating the need for individual connections and allowing for high complexity networking in a small space, with the passband serving as a summing element and enabling weighted signal transmission through modulation of radiation characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional electrical connections are used between nodes, then signal transmission between nodes is achieved, but the device complexity increases quadratically with the number of nodes
Solution Approach 1:
The patent replaces conventional electrical conductor tracks with electromagnetic radiation (optical signals) for signal transmission between nodes. Each node uses transmitting elements (emitters) and receiving elements (detectors) to communicate through electromagnetic fields, eliminating the need for physical electrical connections and reducing the quadratic complexity of wiring.
Solution Approach 2:
The patent implements a shared common passband that serves multiple functions simultaneously: it acts as a transmission medium for electromagnetic radiation, provides signal summation capability, and enables parallel communication between multiple nodes. This multi-functional approach reduces the number of dedicated connections needed.
2Adaptability or versatility
If the number of nodes is increased to enhance network capability, then the processing power of the neural network is improved, but the physical space required and device complexity increase
Solution Approach 1:
By substituting physical electrical connections with electromagnetic radiation transmission, the patent enables high-density node integration without proportionally increasing physical space requirements. The shared common passband allows multiple nodes to communicate simultaneously through the same physical medium, reducing the space needed for interconnections.
3Reliability
If individual optical connections are implemented between nodes, then signal transmission reliability is improved, but the manufacturing complexity and cost increase
Solution Approach 1:
The patent merges multiple individual optical connection functions into a single shared common passband. Instead of implementing separate optical fibers or waveguides for each node pair, all nodes share the same electromagnetic transmission medium, dramatically simplifying manufacturing while maintaining reliable signal transmission through the unified passband structure.
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 approach enables efficient networking of a large number of nodes with reduced physical complexity, allowing for scalable and compact neural network implementations while simplifying the weighting process by integrating signal summation and transmission within the common passband.
Implementation Method 1
the nodes of the first logic layer have transmitting elements configured for emitting electromagnetic radiation and the nodes of the second logic layer have receiving elements configured for receiving electromagnetic radiation
Implementation Method 2
the group of transmitting elements of the first logic layer be coupled with the associated group of receiving elements of the adjacent second logic layer via a common passband that allows electromagnetic radiation to pass through
Implementation Method 3
using a shared passband that receives and transmits all the electromagnetic radiation emitted by the nodes for signal transmission has the advantage of allowing it to act as a summing element. The radiation coupled in by the transmitting elements (i.e., the excitation signals) is aggregated in the shared passband
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention discloses a circuit arrangement for processing signals with a neural network having a plurality of nodes, wherein a number of nodes form a common logic layer, and wherein each node of a group of nodes of a first logic layer is connected to each node of a group of nodes of an adjacent second logic layer, and wherein the nodes of the first logic layer have transmitting elements configured for emitting electromagnetic radiation and the nodes of the second logic layer have receiving elements configured for receiving electromagnetic radiation. The group of transmitting elements of the first logic layer is coupled to the associated group of receiving elements of the second logic layer via a common electromagnetic-permeable passband in order to transmit radiation emitted by each transmitting element to all coupled receiving elements.The transmitting elements are connected to a control unit that defines a characteristic of the emitted radiation by means of a learned weight. A first assignment unit is connected to the control units of the transmitting elements, and a second assignment unit is connected to the receiving elements. The assignment units are configured to assign the signals emitted by the transmitting elements to the receiving elements, which are coupled via the common passband.