Ising Device Arbitration Circuit Parallel Update Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Ising devices face challenges in achieving convergence to an optimal solution when multiple neuron circuits are updated in parallel, leading to potential deterioration in calculation speed and efficiency in solving multivariable optimization problems.
Innovation Solution
An Ising device comprising a plurality of neuron circuits that calculate output values based on weighted connections, with an arbitration circuit controlling updates to ensure only interconnected neuron circuits change simultaneously, while allowing non-interconnected circuits to update concurrently, thereby improving convergence and reducing calculation time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If multiple neuron circuits are updated simultaneously in parallel, then calculation speed is improved, but convergence to optimal solution deteriorates
Solution Approach 1:
The patent segments the neuron circuits into two groups: interconnected neuron circuits and non-interconnected neuron circuits. This segmentation allows different update strategies to be applied to different groups, resolving the contradiction by enabling parallel updates only for non-interconnected circuits while maintaining sequential updates for interconnected circuits to ensure convergence.
Solution Approach 2:
The patent dynamically adjusts the update strategy based on the interconnection relationships between neuron circuits. The arbitration circuit determines in real-time which circuits can be updated in parallel versus sequentially, making the system adaptive to the specific problem structure and maintaining both speed and convergence.
2Productivity
If all neuron circuits are updated in parallel, then processing efficiency is improved, but calculation accuracy deteriorates
Solution Approach 1:
The patent divides neuron circuits into segments based on their interconnection patterns. Non-interconnected circuits are updated in parallel to maintain efficiency, while interconnected circuits are updated sequentially to preserve accuracy. This segmented approach resolves the contradiction between productivity and precision.
Solution Approach 2:
Different update qualities are applied to different parts of the system: parallel updates (higher efficiency, lower precision control) are applied to non-interconnected circuits, while sequential updates (lower efficiency, higher precision control) are applied to interconnected circuits. This local differentiation resolves the global contradiction.
3Reliability
If sequential updates are used for all neuron circuits, then convergence is ensured, but calculation time increases
Solution Approach 1:
The patent segments the update process into sequential updates for interconnected circuits (ensuring convergence) and parallel updates for non-interconnected circuits (reducing time loss). This segmentation resolves the contradiction by applying the appropriate update mode to the appropriate circuit segment.
Solution Approach 2:
Instead of applying sequential updates to all circuits (excessive action that ensures convergence but wastes time), the patent applies sequential updates only where necessary (partial action - only for interconnected circuits), while using parallel updates where safe. This partial application resolves the contradiction between reliability and time loss.
Data Source
AI summary
An individual neuron circuit calculates a first value based on a sum of products each obtained by multiplying one of weight values, each representing connection or disconnection between a corresponding neuron circuit and one of the other neuron circuits, by a corresponding one of output signals of the other neuron circuits and outputs 0 or 1, based on a result of comparison between a second value obtained by adding a noise value to the first value and a threshold. An arbitration circuit allows, when first output signals of first neuron circuits interconnected among the neuron circuits simultaneously change based on the weight values, updating of only one of the first output signals of the first neuron circuits and allows, when second output signals of second neuron circuits not interconnected simultaneously change, updating of the second output signals.