Please enable JavaScript.

Coggle requires JavaScript to display documents.

Ensemble Learning (Introduction (No single model that will always do…

- - - - Base-learners are trained using
        different training sets
      - Learners are aggregated using
        their average or a voting system
      - Different training sets are
        generated by bootstrapping
    - - Utilize weighted averages to make
        weak learners into strong learners
      - More weights are given to datasets
        with the most misclassified samples
    - - Combining the predictions from
        base-learners by voting
      - All learners are given weights, \(w_j \)
        , where j = classifier
      - Simple voting
        
        All classifiers have equal weight
      - More weights to more accurate classifiers
      - Majority Voting
        \( y_k = w_j d_{jk} \)
        , where j = classifier
        k = class
        
        Final output prediction is the one that
        receives more than half of the votes
        
        If NO => ensemble method could not make
        a stable prediction for this instance