What is the basic difference between a Perceptron and a Naive Bayes classifier?

         Perceptron          |          Naive Bayes
(1) Perceptron uses Neural-  |(1) Naive Bayes uses probabi-
    network for learning and |    listic theory for learning
    classification.          |    and classification
(2) Perceptron reads one sa- |(2) Naive Bayes needs to read-
    mple at a time to update |    the entire training data 
    its knowledge about the  |    before updating its knowl-
    training data. This is   |    edge about the training 
    called online learning.  |    data.
(3) In case of Perceptrons,  |(3) Training and test data are  
    training-data also serve |    different.
    the purpose of test data |

what more differences do they have?

  • $\begingroup$ Why do you need more differences?:) $\endgroup$ – Diligent Key Presser Dec 31 '16 at 5:07
  • 3
    $\begingroup$ @DiligentKeyPresser, (1) I don't know if my provided differences are correct or or, (2) to know the difference between Naive Bayes and Off-line Percepton. $\endgroup$ – user3642 Dec 31 '16 at 5:11

There is different idea behind them...

Naive Bayes is based on some reach background of Probability Theory... It tries to find a "Theory" that is consistent with "Observations" by using the Bayes Theorem. But technically it will be so simplified to be applied. Actually you are solving some kind of Optimization which Statisticians do. You need all of data at once.

But Perceptron are more heuristic. The Math behind it, is some kind poor. But it works in reality. You takes your data (at once or releasing during time) and try to change the weights iteratively hopefully to find a good network... The idea is so simple. In this case, indeed, you are going to solve an Optimization problem but objective function and the method is compeletely different

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy