Sunday, 12 April 2015

MOA: Massive Online Analysis:
                         Massive Online Analysis is a framework for online learning from data streams. It is closely related to WEKA It includes a collection of offline and online as well as tools for evaluation: boosting and bagging Hoeffding Trees with and without Naïve Bayes classifiers at the leaves.
 Data stream classification cycle:
 1 Process an example at a time, and inspect it only once (at most)
 2 Use a limited amount of memory
3 Work in a limited amount of time
4 Be ready to predict at any point.




Classification Experimental setting:
 Evaluation procedures for Data Streams
                  Ø  Holdout
                  Ø  Interleaved Test-Then-Train or Prequential
Environments
Ø  Sensor Network: 100Kb
Ø   Handheld Computer: 32 Mb
Ø   Server: 400 Mb
  Data Sources
      Ø  Random Tree Generator
      Ø   Random RBF Generator
      Ø   LED Generator
      Ø  Waveform Generator
      Ø  Hyperplane SEA Generator
      Ø   STAGGER Generator
 Classifiers
                   Ø  Naive Bayes
                   Ø   Decision stumps
                   Ø  Hoeffding Tree
                   Ø  Hoeffding Option Tree
                   Ø  Bagging and Boosting 
                   Ø  ADWIN Bagging and Leveraging Bagging
Prediction strategies
                    Ø  Majority class
                    Ø   Naive Bayes Leaves

                    Ø  Adaptive Hybrid 

No comments:

Post a Comment