Using ELM-based weighted probabilistic model in the classification of synchronous EEG BCI

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Extreme Learning Machine (ELM) offers efficient machine learning with fast implementation.
• Classifying electroencephalography (EEG) signals is crucial for brain-computer interface (BCI) systems.
• Existing methods require improvement in accuracy and efficiency for BCI applications.

Purpose of the Study:

• To propose a novel hybrid method combining ELM with a probabilistic model for EEG signal classification.
• To enhance the discriminative performance of BCI systems through cumulative learning.
• To evaluate the proposed method against established classification techniques.

Main Methods:

• A hybrid model integrating Extreme Learning Machine (ELM) with a probabilistic approach.
• Utilizing the softmax function for converting ELM outputs to classification probabilities.
• Incorporating the Chernoff error bound from Bayesian models as a discriminant weight during training.

Main Results:

• The proposed method demonstrated cumulative performance improvement by leveraging all preceding training data.
• Comparative analysis on BCI competition datasets showed competitive results against Linear Discriminant Analysis, Support Vector Machine, ELM, and weighted probabilistic models.
• Evaluation metrics included mutual information, classification accuracy, and information transfer rate.

Conclusions:

• The novel ELM-probabilistic model hybrid method offers a competitive and effective approach for EEG signal classification in BCI.
• The cumulative learning strategy enhances discriminative capabilities.
• This approach shows promise for advancing BCI technology through improved signal processing.