Neuromorphic Resistive Network Correction Circuitry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neuromorphic computational systems face computational inaccuracies due to finite ON/OFF conductance ratios in non-volatile resistive devices, which are not adequately addressed by assuming infinite ratios, leading to significant errors in applications like image recognition.
Innovation Solution
The implementation of a cross-point resistive network with line control circuitry, where variable resistive units generate correction and resultant line currents to correct for the finite ON/OFF conductance ratios, allowing for improved computational accuracy by adjusting digital vector values based on these currents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If non-volatile resistive devices are used in neuromorphic computational systems, then device compactness and cost are improved, but computational accuracy deteriorates due to finite ON/OFF conductance ratios
Solution Approach 1:
A correction column is introduced as an intermediary component between the data columns and the read circuitry. This correction column generates compensation currents that mediate the effect of finite ON/OFF ratios, allowing the system to use compact non-volatile resistive devices while maintaining computational accuracy through current correction
2Device complexity
If the ON/OFF conductance ratio is assumed to be infinite, then computational complexity is reduced, but computational accuracy deteriorates due to the non-zero off conductance state
Solution Approach 1:
The system changes the parameter representation by separating the data storage function (in data columns) from the correction function (in correction columns). By modifying how conductance states are interpreted and compensated, the system achieves accurate computation without requiring infinite ON/OFF ratios, thus maintaining simplicity while improving accuracy
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 significantly reduces or eliminates computational errors caused by finite ON/OFF ratios, enhancing the accuracy and reliability of neuromorphic computational systems, particularly in tasks such as image recognition.
Implementation Method 1
The cross point resistive network has variable resistive units and a set of conductive lines. Sets of the variable resistive units are connected to a corresponding conductive line. One of the sets of the variable resistive units is configured to generate a correction line current along its corresponding conduction line
Data Source
AI summary
Neuromorphic computational circuitry is disclosed that includes a cross point resistive network and line control circuitry. The cross point resistive network includes variable resistive units. One set of the variable resistive units is configured to generate a correction line current on a conductive line while other sets of the variable resistive units generate resultant line currents on other conductive lines. The line control circuitry is configured to receive the line currents from the conductive lines and generate digital vector values. Each of the digital vector values is provided in accordance with a difference between the current level of a corresponding resultant line current and a current level of the correction line current. In this manner, the digital vector values are corrected by the current level of the correction line current in order to reduce errors resulting from finite on to off conductance state ratios.