Comparison of Extreme Learning Machine and K-Nearest Neighbour Performance in Classifying EEG Signal of Normal, Poor
Summary
Extreme learning machine (ELM) accurately identified dyslexic children using EEG signals during writing tasks. This machine learning approach shows promise for recognizing differences in brain processing related to dyslexia.
Area of Science:
- Neuroscience
- Machine Learning
- Developmental Psychology
Background:
- Dyslexia is a learning difficulty impacting number and letter processing.
- Electroencephalogram (EEG) analysis offers insights into brain processing differences.
- Distinguishing between normal, poor, and capable dyslexic children is crucial.
Purpose of the Study:
- To compare the effectiveness of two machine learning techniques in classifying EEG signals of dyslexic children.
- To evaluate the performance of k-nearest neighbour (KNN) and extreme learning machine (ELM) for dyslexia detection.
- To determine the reliability of EEG signal analysis in identifying different levels of dyslexia.
Main Methods:
- Applied k-nearest neighbour (KNN) with correlation distance and extreme learning machine (ELM) with radial basis function (RBF).
- Utilized EEG signals recorded during writing of words and non-words from normal, poor, and capable dyslexic children.
- Assessed classifier performance using sensitivity, specificity, and accuracy metrics.
Main Results:
- Extreme learning machine (ELM) achieved 89% accuracy in classifying dyslexic children.
- K-nearest neighbour (KNN) achieved 83% accuracy.
- ELM demonstrated superior performance compared to KNN in this classification task.
Conclusions:
- Extreme learning machine (ELM) is a feasible and reliable method for recognizing normal, poor, and capable dyslexic children.
- EEG signal analysis combined with ELM can effectively differentiate dyslexic individuals.
- This approach holds potential for aiding in the diagnosis and understanding of dyslexia.
