Does patternnet create MLP neural network??
3 ビュー (過去 30 日間)
古いコメントを表示
I want to create a NN to classify Iris Data set with a specific algorithm like (ABC) this NN should be MLP-NN I see a lot of questions and their answer but I can't really consider if patternnet creates MLP-nn or not can any body ensure me???
0 件のコメント
採用された回答
Greg Heath
2018 年 1 月 4 日
Yes. The configuration generated by the call to patternnet is a
Multi-layer-perceptron
https://en.wikipedia.org/wiki/Perceptron
Thank you for formally accepting my answer
Greg
2 件のコメント
Walter Roberson
2018 年 1 月 4 日
The documentation at https://www.mathworks.com/help/nnet/ref/perceptron.html specifically says that patternnet does nonlinear separation and that perceptrons never do.
Greg Heath
2018 年 1 月 5 日
編集済み: Greg Heath
2018 年 1 月 5 日
Quite a bit of confusion occurs because there is a lack of understanding concerning the term "perceptron" because
The single term perceptron DOES NOT IMPLY HIDDEN LAYERS.
The acronym MLP implies a multilayer perceptron.
Therefore one has to read carefully to be sure which one the referral concerns.
The default configurations of fitnet and patternet have a single hidden layer and are, therefore, MLPs. HOWEVER, the default of 10 can be overwritten to 0. Then the configuration becomes a perceptron.
Hope this helps. (Yeah, I know!)
Greg
その他の回答 (1 件)
Walter Roberson
2018 年 1 月 3 日
No, patternnet does not use MLP.
and see the File Exchange for a number of MLP contributions.
5 件のコメント
Greg Heath
2018 年 1 月 5 日
編集済み: Walter Roberson
2018 年 1 月 5 日
- ONE hidden layer is sufficient.
- Use FITNET for curveFITting and regression.
- Use PATTERNNET for PATTERN-recognition and classification.
- "F"eed"F"orward net new"ff" and special cases new"fit" for curve"FIT"ting and new"pr" for "P"attern "R"ecognition are obsolete.
- 10 neurons in a hidden layer is a default that does not have to be specified
- The basic code for each is given in the help and doc documentation:
help fitnet
doc fitnet
and similarly for patternnet.
In addition, I have posted zillions of examples, including tutorials, in both the NEWSGROUP and ANSWERS.
参考
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!