Neuromorphic Computer Architecture Using Variable Resistance Circuits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current software-based neuromorphic computing methods are impractical for everyday use due to their high requirements for high-performance computers and lack of hardware implementations that mimic brain-like synaptic and neuron functionality, making them unsuitable for mobile and low-cost systems.
Innovation Solution
A neuromorphic computer architecture utilizing electronic devices with variable resistance circuits to represent synaptic connection strength and positive/negative output circuits to mimic excitatory and inhibitory responses, enabling high-density fabrication of brain-like computing functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If software-based neuromorphic computing is used, then brain-like computing functions can be implemented, but high-performance computers and super computers are required making it impractical for mobile and low-cost systems
Solution Approach 1:
The patent replaces software-based neuromorphic computing with a hardware-based electronic system that uses electronic devices to directly mimic neuron and synapse behavior. This substitution of software with specialized hardware eliminates the need for high-performance computers while achieving brain-like computing functions in compact, low-cost systems.
Solution Approach 2:
The patent changes the operational parameters by using electronic devices with variable resistance to represent synaptic weights and threshold voltages to represent neuron firing thresholds. This parameter-based approach allows direct hardware implementation of neuromorphic functions without requiring complex software processing on powerful computers.
2Adaptability or versatility
If hardware-based neuromorphic computer is designed with large number of neurons and synaptic connections, then practical brain functionality can be achieved, but electronic devices must be small to be fabricated within small physical area
Solution Approach 1:
The patent segments the neuromorphic system into discrete electronic devices where each device represents a neuron or synapse. This segmentation allows the system to be fabricated using standard semiconductor manufacturing processes, enabling large numbers of neurons and connections to be packed into small areas through integrated circuit technology.
Solution Approach 2:
The patent merges multiple functions into single electronic devices - using the same electronic components to simultaneously represent synaptic weights, transmit signals, and implement neuron activation functions. This functional merging reduces the physical area required compared to dedicated components for each function.
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 allows for the implementation of human brain-like computing in compact electronic systems, overcoming the limitations of software-based methods and enabling practical applications such as image recognition by mimicking the brain's synaptic and neuron behavior.
Implementation Method 1
utilizing electronics to perform the function of neurons and synaptic connections. The invention provides variable resistance circuits to represent interconnection strength between neurons
Implementation Method 2
a positive and negative output circuit to represent excitatory and inhibitory responses, respectively
Data Source
AI summary
A neuromorphic computing device utilizing electronics to perform the function of neurons and synaptic connections. The invention provides variable resistance circuits to represent interconnection strength between neurons and a positive and negative output circuit to represent excitatory and inhibitory responses, respectively. The invention provides advantages over software-based neuromorphic computing methods.