Variable-Resistance Information Processing for Programmable Short-Term Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Designing short term memory property in neural networks substituted with physical elements is difficult, which affects the performance of tasks requiring consideration of past data in time series analysis.
Innovation Solution
An information processing device is implemented using physical elements with variable resistance or conductance, where input signals and control signals adjust the memory properties of these elements to optimize short term memory, allowing for hardware realization of neural network operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If neural network operations are performed using mathematical calculations, then processing accuracy can be maintained, but processing efficiency is insufficient due to high arithmetic load
Solution Approach 1:
The patent replaces mathematical calculation operations with physical electrical operations using resistive elements. The neural network computations are performed through voltage division and current flow in resistor networks, substituting the mechanical/electronic computational process with a purely electrical physical process that operates faster and with lower energy consumption while maintaining computational accuracy.
Solution Approach 2:
The patent changes the operational parameters from digital mathematical computations to analog electrical parameters (voltage, current, resistance). By representing neural network weights and inputs as electrical resistance values and voltages, the system performs computations through physical electrical laws (Ohm's law, Kirchhoff's laws) rather than mathematical algorithms, achieving both speed and accuracy.
2Productivity
If physical elements are used to substitute mathematical operations, then processing efficiency improves, but it becomes difficult to design short term memory property
Solution Approach 1:
The patent introduces dynamic control capabilities to the resistive elements, allowing their resistance values to be changed over time based on control signals. This dynamic adjustability enables the physical elements to exhibit temporal behavior patterns, including short-term memory properties, by modulating resistance values in response to input sequences and time-dependent control inputs.
Solution Approach 2:
The patent utilizes time-varying parameter changes in the resistive elements to implement short-term memory functionality. By dynamically adjusting resistance values based on temporal patterns of input signals and control voltages, the system can retain information for specific durations and forget it afterward, creating programmable short-term memory behavior in the physical hardware.
3Reliability
If all arithmetic operations are mathematically performed, then correct answer rate can be maximized, but operational load remains high
Solution Approach 1:
The patent replaces energy-intensive mathematical arithmetic operations with low-power electrical operations in passive resistor networks. Instead of active computational processors performing multiplications and additions, the system uses passive electrical components where computations emerge naturally from voltage and current distributions, dramatically reducing operational energy consumption while maintaining computational correctness.
Solution Approach 2:
The patent implements self-computing electrical circuits where the resistor network automatically performs neural network computations through physical electrical laws without requiring active control or power consumption during the computation phase. The system leverages the natural behavior of electrical currents and voltages to perform calculations, eliminating the need for energy-intensive active processing.
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 device can freely design and adjust short term memory properties, enhancing the correct answer rate for time series data analysis by maintaining past information for a controlled duration, reducing operational load.
Implementation Method 1
a physical element of which resistance or conductance changes in accordance with application of a voltage or a current
Data Source
AI summary
An information processing device according to the present disclosure includes an input unit, an information processing unit, and an output unit. The information processing unit includes a physical element of which resistance or conductance changes in accordance with application of a voltage or a current. The input unit is configured to be able to input an input signal and a control signal to the physical element. The input signal is a signal for changing the resistance or the conductance of the physical element on the basis of input information. The control signal is a signal for changing a change speed of the resistance or the conductance of the physical element. The output unit is configured to receive a signal from the physical element and output an output signal.