Natural language processing using knn
By performing similarity operations in parallel on associative memory devices, the problem of low sorting efficiency of the k nearest neighbor algorithm on large datasets is solved, enabling fast searching of k minimum values in constant time and improving data processing efficiency.
CN117370563BActive Publication Date: 2026-05-29GSI TECHNOLOGY INC
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GSI TECHNOLOGY INC
- Filing Date
- 2018-07-16
- Publication Date
- 2026-05-29
Smart Images
Figure CN117370563B_ABST
Abstract
A system for natural language processing includes a memory array and a processor. The memory array is partitioned into a similarity section that stores a plurality of feature vectors, a SoftMax section for determining a probability of occurrence of a feature vector, a value section that stores a plurality of modified feature vectors, and a tag section. The processor activates the array to perform, in parallel, in each column indicated by the tag section: a similarity operation in the similarity section between a vector question and a feature vector stored in the indicated column; a SoftMax operation in the SoftMax section for determining an associated SoftMax probability value for the indicated feature vector; a multiplication operation in the value section for multiplying the associated SoftMax value by a modified feature vector stored in the indicated column; and a vector sum in the value section for accumulating a vector of interest of outputs of the multiplication operation.
Need to check novelty before this filing date? Find Prior Art