Synapse Memory Cell Structure for Neuromorphic Weight Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current synapse memory cells, particularly analog and Multi-Level Cell (MLC) types, face challenges in accurately controlling synapse weight values, requiring complex peripheral circuits and high-resolution sensing circuits, which complicates their implementation in neuromorphic systems.
Innovation Solution
A synapse memory cell structure utilizing a plurality of unit resistances connected in parallel, allowing for a simple and accurate representation of synapse weight values using a first set of states dependent on a second set, eliminating the need for high-resolution sensing circuits and complicated encoding/decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If analog or MLC synapse memory cells are used to achieve high precision synapse weight representation, then measurement precision is improved, but device complexity increases due to complex peripheral circuits and high-resolution sensing circuits
Solution Approach 1:
The synapse weight value is segmented into multiple discrete levels (first set of states) that can be represented by combinations of simpler binary states (second set of states). Each cell component stores one level of the segmented weight value, and multiple cell components work together to represent the complete synapse weight, thereby achieving high precision without requiring complex single-cell structures
Solution Approach 2:
Multiple cell components are merged to form a single synapse memory cell. Each cell component uses simple binary states (second set), but their combined output represents more complex synapse weight values (first set). This merging allows the system to achieve high measurement precision through multiple simple units working together rather than requiring complex individual cells
2Productivity
If MLC synapse memory cells are used to increase storage capacity, then productivity is improved, but manufacturing precision requirements increase
Solution Approach 1:
The synapse weight value is segmented into multiple discrete levels (first set of states) that can be represented by combinations of simpler binary states (second set of states). Each cell component stores one level of the segmented weight value, and multiple cell components work together to represent the complete synapse weight, thereby achieving high precision without requiring complex single-cell structures
Solution Approach 2:
The system changes the parameter representation from requiring high-precision control of a single analog value to controlling multiple discrete binary states. By transforming the control parameter from continuous analog voltage to discrete digital states, the manufacturing precision requirement is relaxed while maintaining high storage capacity and weight representation accuracy
3Measurement precision
If high-resolution sensing circuits are used to accurately read synapse weights, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The synapse weight value is segmented into multiple discrete levels (first set of states) that can be represented by combinations of simpler binary states (second set of states). Each cell component stores one level of the segmented weight value, and multiple cell components work together to represent the complete synapse weight, thereby achieving high precision without requiring complex single-cell structures
Solution Approach 2:
The system replaces the need for complex high-resolution sensing circuits with a simpler digital readout approach. Instead of using complex analog sensing to directly measure continuous weight values, the system uses simple sensing to read discrete binary states from multiple cell components, which are then digitally combined to reconstruct the full synapse weight value with high precision
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 enables precise control of synapse weight values without the need for complex circuits, improving the accuracy and convenience of synapse memory cells in neuromorphic systems by using a straightforward resistive memory cell structure.
Implementation Method 1
a first resistance value corresponding to a first weight value and a second resistance value corresponding to a second weight value
Data Source
AI summary
A memory cell structure includes a plurality of write lines arranged for writing a synapse state to a synapse memory cell including a plurality of cell components each including at least one unit cell, each of the plurality of write lines being used for writing the synapse state by writing a first set of states to a corresponding cell component of the plurality of cell components by writing one of a second set of states to each unit cell included in the corresponding cell component, and the first set depending on the second set.