151. Integer Sparse Distributed Memory Based on Neural Coding
- Author
-
Xiaowei Wang and Hongying Meng
- Subjects
Theoretical computer science ,Integer ,Data retrieval ,sparse distributed memory ,Computer science ,Encoding (memory) ,Pattern recognition (psychology) ,SDM ,neural coding ,Memory model ,Sparse distributed memory ,Neural coding ,Associative property - Abstract
Sparse Distributed Memory (SDM) is a mathematical associative human long-term memory model that is psychologically and neuroscientifically plausible. It is content addressable and can potentially store unlimited high dimensional information although it only accepts binary vectors. Although an integer SDM has been proposed to store non-binary vectors, the size of the model has been increased significantly. In this paper, we propose a biologically plausible encoding scheme to enable SDM to accept integer vectors, requiring much less space than the existing integer SDM model. A public handwritten digits dataset with integer values has been used for the evaluation of our proposed model and satisfied performance has been achieved in both data retrieval and pattern recognition.
- Published
- 2021