Neuromorphic Crossbar Array Storing Weights and Neuronal States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neuromorphic devices based on crossbar array structures require external digital circuitry to implement neuron functionality, limiting their efficiency and requiring conversion of signals from analog to digital domains.
Innovation Solution
A neuromorphic device with a crossbar array structure and an integrated analog circuit that stores synaptic weights and neuronal states using memristive devices, allowing computations to be performed directly in the analog domain without additional circuitry, enabling efficient computation of neuronal states and outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If external digital circuitry is used to implement neuron functionality, then the device can perform computational tasks, but the device complexity increases and signal conversion between analog and digital domains is required
Solution Approach 1:
The patent merges the functions of synaptic weight storage and neuronal state computation into a single crossbar array structure. The crossbar array simultaneously stores synaptic weights in its conductance values and performs neuronal state computations through analog signal processing, eliminating the need for separate digital circuitry and reducing overall device complexity.
Solution Approach 2:
The crossbar array is designed to perform multiple functions: it stores synaptic weights, computes neuronal states through analog multiplication and accumulation, and generates output signals. This multi-functional design replaces what would traditionally require separate digital circuits, reducing device complexity while maintaining computational versatility.
2Adaptability or versatility
If external digital circuitry is used to implement neuron functionality, then the device can perform computational tasks, but additional signal conversion from analog to digital domains is required
Solution Approach 1:
The patent combines analog signal processing directly within the crossbar array structure, allowing neuronal state computations to be performed in the analog domain without requiring conversion to digital. This eliminates the time loss associated with analog-to-digital conversion while maintaining the ability to implement complex neuron functionalities.
3Productivity
If additional circuitry is used for neuron functionality, then the device can perform computational tasks, but the device complexity and resource requirements increase
Solution Approach 1:
The patent merges computation and storage functions into the crossbar array structure itself. The same hardware components that store synaptic weights also perform the computational operations, eliminating the need for additional dedicated circuitry and reducing device complexity while maintaining high computational efficiency.
Solution Approach 2:
The crossbar array performs computations using its own stored data without requiring external assistance. The synaptic weights stored in the crossbar array are directly used in the analog multiplication and accumulation operations, allowing the device to serve itself and eliminating the need for additional control circuitry.
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 enhances computational efficiency by eliminating the need for digital conversion and additional circuitry, allowing for faster and more direct computation of neuronal states and outputs, particularly suitable for both standard and spiking neural networks.
Implementation Method 1
The crossbar array structure comprises N×M electronic devices, which, in preferred embodiments, include, each, a memristive device
Data Source
AI summary
Neuromorphic methods, systems and devices are provided. The embodiment may include a neuromorphic device which may comprise a crossbar array structure and an analog circuit. The crossbar array structure may include N input lines and M output lines interconnected at junctions via N×M electronic devices, which, in preferred embodiments, include, each, a memristive device. The input lines may comprise N1 first input lines and N2 second input lines. The first input lines may be connected to the M output lines via N1×M first devices of said electronic devices. Similarly, the second input lines may be connected to the M output lines via N2×M second devices of said electronic devices. The analog circuit may be configured to program the electronic devices so as for the first devices to store synaptic weights and the second devices to store neuronal states.