Memory Device Vectoring Search Data Using Discrete Cell States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing similarity analysis methods in fields like text mining and human face recognition face challenges in achieving simple calculations with high analysis confidence, as they often require complex vector comparisons and high computational resources.
Innovation Solution
A memory device and data search method that compresses search data and database objects using a recorded compression mode, generating search data vectors to determine matches based on Hamming distance, which reduces dimensionality and improves computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If complex vector comparisons are used for similarity analysis, then analysis confidence is improved, but computational resources and calculation complexity increase
Solution Approach 1:
The patent transforms the similarity analysis from continuous vector comparison to discrete memory cell state comparison by changing the parameter representation. Feature vectors are encoded into memory cell threshold voltages with discrete states (erased, first state, second state), converting continuous similarity measurements into discrete, easily comparable states that reduce computational complexity while maintaining analysis confidence
Solution Approach 2:
The patent replaces complex software-based vector comparison algorithms with hardware-based memory cell state comparison. The similarity analysis is performed through physical memory operations (reading and comparing memory cell states) rather than computational vector operations, significantly reducing calculation complexity and improving processing speed
2Measurement precision
If high-dimensional vectors are used for accurate similarity analysis, then analysis precision is improved, but storage requirements and processing time increase
Solution Approach 1:
The patent segments the high-dimensional feature vector into multiple discrete components, each represented by memory cell states. The vector encoding divides the continuous vector space into discrete regions represented by different memory cell states (erased, first state, second state), enabling parallel comparison of multiple dimensions simultaneously through memory array operations, thus reducing processing time while maintaining precision
Solution Approach 2:
The patent adds a temporal dimension to the storage by using multi-state memory cells that can represent multiple vector dimensions in a single cell through different threshold voltage states. This allows the system to store and process high-dimensional vectors more efficiently by utilizing the voltage state dimension rather than requiring separate storage locations for each vector component
3Measurement precision
If traditional vector comparison methods are used, then analysis accuracy is maintained, but computational efficiency decreases
Solution Approach 1:
The patent enables the memory device to perform similarity analysis autonomously through hardware-based operations. The memory cells themselves participate in the comparison process by maintaining their states representing vector components, and the comparison is performed through direct memory read operations and state matching, eliminating the need for external computational processing and significantly improving computational efficiency
Solution Approach 2:
The patent creates a hardware copy of the vector data in memory cell states that can be rapidly compared against query vectors. The feature vectors are encoded into memory cell threshold voltage distributions that serve as searchable copies, enabling fast similarity searches without repeatedly processing the original high-dimensional vector data, thus improving computational efficiency while maintaining 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 enables fast and accurate similarity searches with reduced computational resources and storage requirements, enhancing analysis confidence and efficiency in similarity analysis tasks.
Implementation Method 1
n memory cells among the m memory cells are programmed to have a first threshold voltage and (m-n) memory cells among the m memory cells are programmed to have a second threshold voltage
Implementation Method 2
based on a current sensing result on the first bit line, determining whether the search data is matched with the first feature vector of the first object stored in the memory cell group
Data Source
AI summary
A data search method for a memory device is provided. The data search method includes: based on a recorded compression mode, vectoring a search data to generate a search data vector, and based on the recorded compression mode, compressing the search data and a plurality of objects in a database; setting a search condition; searching the objects of the database by the search data vector to determine whether the search data is matched with the objects of the database; and recording and outputting at least one matched object of the database, the at least one matched object matched with the search data.