Eeg Analysis For Short Term Memory Modeling In Visually Explored Shape Recognition Tasks
2017 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (SSCI)(2017)
摘要
The paper introduces a novel approach to short term memory modeling using electroencephalographic signals acquired from occipital, prefrontal and parietal lobes of the subjects during the recognition of asymmetric two-dimensional object shapes. Each subject is presented with a two-dimensional object of irregular geometry and is asked to remember it for a short period until he/she is instructed to draw the two-dimensional shape of the object. After the drawing is complete, the reference two-dimensional shape and the one produced by the subject are compared.The paper attempts to model Short term memory using unsupervised Hebbian Learning, considering functional invariance of the other relevant brain modules lying on the encoding and recall pathways of the memory. An error norm is defined to measure the magnitude of the error between the input and reproduced shape pattern, and thereby is used to adapt the short term memory. A small value in error indicates proper learning of the proposed short term memory model, and thus can be used to discriminate people with memory failure. Experiments undertaken reveal that the error metric could be used successfully to detect memory failures in five patients, two of whom were in early stages of Alzheimer's disease, two of whom suffer from amnesia, and one patient with parietal lobe damage. The proposed short term memory model thus is expected to have applications in detecting memory failures in shape-recognition tasks.
更多查看译文
关键词
short term memory, iconic memory, Hebbian learning, back propagation neural network, shape recognition, average gamma power, memory failure
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络