Spintronic Resonator Synaptic Chain for High-Density Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural networks face limitations in performance due to the large size of neurons and synapses, which restricts the number that can be integrated on a limited surface such as a microchip, leading to decreased network performance.
Innovation Solution
A synaptic chain for neural networks is proposed, where each synapse is a spintronic resonator electrically connected in series by a transmission line, allowing for a greater number of neurons and synapses to be integrated while maintaining a compact footprint.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If neurons and synapses are implemented using conventional CMOS or optical technologies, then each neuron and synapse occupies several tens of square micrometers, but the number of neurons and synapses that can be integrated on a limited surface is limited
Solution Approach 1:
The patent transitions from planar integration to three-dimensional stacking by placing spintronic resonators in vertical stacks of superimposed layers. This allows multiple resonators to occupy the same footprint area at different heights, dramatically increasing the number of neurons and synapses that can be integrated on a limited surface without increasing the chip area.
Solution Approach 2:
The patent implements nesting by placing multiple spintronic resonators within a single footprint area through vertical stacking. Each resonator is formed as a stack of superimposed layers, effectively nesting multiple functional elements within the same spatial envelope, thereby increasing integration density.
2Productivity
If the number of neurons and synapses is increased to improve network performance, then the computational capability increases, but the integration density on limited surface decreases
Solution Approach 1:
By utilizing the vertical dimension through stacked layers, the patent achieves high integration density while maintaining large numbers of neurons and synapses. The three-dimensional architecture allows the network to scale up computational capability without proportionally increasing chip area, thus improving both performance and integration density simultaneously.
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
The proposed synaptic chain architecture enables the creation of neural networks with improved performance by increasing the number of neurons and synapses, while also facilitating fast, low-power, and real-time learning capabilities.
Implementation Method 1
each resonator being suitable for generating between the terminals a direct voltage whose amplitude depends on the deviation of the resonance frequency of the resonator from a reference frequency
Implementation Method 2
each resonator having a resonance frequency
Data Source
AI summary
The invention relates to a synaptic chain of neural networks, the synaptic chain comprising synapses, each synapse being a spintronic resonator, the spintronic resonators being electrically connected in series by a transmission line and being alternately connected.