a b/README.md 

  1
  
# EEG Classifier based on DEAP database





  
  

  2
  





  
  

  3
  
## This repo only include code to proceed DEAP data, no data from DEAP contained!





  
  

  4
  





  
  

  5
  
For access of DEAP dataset, please sign EULA and send a request to DEAP team: http://www.eecs.qmul.ac.uk/mmv/datasets/deap/download.html





  
  

  6
  





  
  

  7
  
## Dependency





  
  

  8
  
* python >= 3.5





  
  

  9
  
* numpy





  
  

  10
  
* pyEEG: https://github.com/forrestbao/pyeeg, need manual fix of source code, refers to issue: https://github.com/forrestbao/pyeeg/issues/26





  
  

  11
  
* scikit-learn





  
  

  12
  
* tensorflow-gpu





  
  

  13
  





  
  

  14
  
## ERD/ERS analysis





  
  

  15
  
Fast Fourier Transformation:





  
  

  16
  
* Windows size = 2 sec





  
  

  17
  
* Windows step = 0.125 sec





  
  

  18
  





  
  

  19
  
## Model attempted





  
  

  20
  
* Support Vector Machine





  
  

  21
  
* Adaboost





  
  

  22
  
* Random Forest





  
  

  23
  
* Artificial neural network





  
  

  24
  





  
  

  25
  
## Best AC rate of Classification on Arousal, Valence, Domaince, Like dimension





  
  

  26
  
* Arousal/Not: 82.6%





  
  

  27
  
* Valence/Not: 83.6%





  
  

  28
  
* Domaince/Not: 81.9%





  
  

  29
  
* Like/Not: 85.1%