Loading...

Application of Evolutionary Algorithms for Effective Feature Selection in Brain Signal Classification in Order to Investigate Memory

Entezari, Saeed | 2015

392 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47750 (55)
  4. University: Sharif University of Technology, International Campus, Kish Island
  5. Department: Science and Engineering
  6. Advisor(s): Shamsollahi, Mohammad Baghe
  7. Abstract:
  8. In order to retrieve the temporal dynamics of the long term memory representation, which is believed that synaptically encoded with respect to an approach, we have utilized machine learning methods to classify the magnetoencephalogram (MEG) data that has been collected from an experiment called association task. The decoding process can be considered as a two-class classification problem in which we want to make a decision about the color or orientation of the grating of the test label. For the first step, different features have been extracted. This extraction can be done from the total signals of the different channels or can be extracted from the signal segments for each channel. These different strategies are used to make an investigation over the importance of different channels in different time intervals. As the number of extracted features for each data point is surprisingly high, we face the problem of curse of dimensionality for the classification process. In this thesis we have used feature selection, which is known as an optimization problem, to reduce the dimension of feature space in order to make the classification process more efficient. In other words, through the selection process, we intend to clarify which channels and time intervals are informative for data representation. We will exploit a binary version of seven different EAs with different individual concepts and criterion to make a selection through different extracted features. Then the classification process will be conducted. The employed strategy for feature selection and voting process contributes to an 8% increase in the classification of the test data
  9. Keywords:
  10. Evolutionary Algorithm ; Classification ; Long-term Memory

 Digital Object List

 Bookmark

No TOC