|
Size: 474
Comment: typo
|
← Revision 5 as of 2009-09-20 23:16:58 ⇥
Size: 476
Comment: converted to 1.6 markup
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 1: | Line 1: |
| Perceptron is a rule-weighting algorithm based on artificial neural networks ("perceptrons") instead of ["GeneticAlgorithm"]s. It is based on error back propagation by gradient descent and includes several security and accuracy-related modifications. It has replaced the GA in the RescoringProcess for SpamAssassin 3.0. | Perceptron is a rule-weighting algorithm based on artificial neural networks ("perceptrons") instead of [[GeneticAlgorithm]]s. It is based on error back propagation by gradient descent and includes several security and accuracy-related modifications. It has replaced the GA in the RescoringProcess for SpamAssassin 3.0. |
| Line 3: | Line 3: |
| Some info may be found at [http://www.spamassassin.org/full/3.0.x/dist/masses/README.perceptron masses/README.perceptron] and at RunningPerceptron. | Some info may be found at [[http://www.spamassassin.org/full/3.0.x/dist/masses/README.perceptron|masses/README.perceptron]] and at RunningPerceptron. |
Perceptron is a rule-weighting algorithm based on artificial neural networks ("perceptrons") instead of GeneticAlgorithms. It is based on error back propagation by gradient descent and includes several security and accuracy-related modifications. It has replaced the GA in the RescoringProcess for SpamAssassin 3.0.
Some info may be found at masses/README.perceptron and at RunningPerceptron.
