The Netlab toolbox is designed to provide the central tools necessary for the simulation of theoretically well founded neural network algorithms and related models for use in teaching, research and applications development. It contains many techniques which are not yet available in standard neural network simulation packages.
The principles behind the toolbox are more important than simply compiling lists of algorithms. Data analysis and modelling methods should not be used in isolation; all parts of the toolbox interact in a coherent way, and implementations of standard pattern recognition techniques (such as linear regression and K-nearest-neighbour classifiers) are provided so that they can be used as benchmarks against which more complex algorithms can be evaluated. This interaction allows researchers to develop new techniques by building on and reusing existing software, thus reducing the effort required and increasing the robustness and usability of the new tools.
An accompanying text book, Netlab: Algorithms for Pattern Recognition written by Ian Nabney is published by Springer in their series Advances in Pattern Recognition: the ISBN number is 1-85233-440-1. More information can be found at http://www.ncrg.aston.ac.uk and http://www.mathworks.com/support/books/book4368.jsp.
It very useful for me and I setup it to appling my project. Thanks author for everything.
I used this toolbox successfully with Matlab R2013a. It is very beneficial to have the book referenced in the above description. My application was classification of sounds in a trained NN into one of several categories. here are a few notes from my specific application:
Activation Functions Investigated
Linear – simplest, gives good results
Softmax – best general purpose for 1 of N classification
Logistic – good for binary classifications
Conjugate gradient descent – worst performing method
Scaled conjugate gradient descent (SCG) – sometimes superior
Quasi-Newton – gives most consistent results for current data set
Search for best number of hidden units
Smaller number runs faster/simpler
Larger number may provide more accurate results with the possibility of over-fitting the available data
Current data set, with 4 possible sound classifications, gave best result with about 15 hidden units
I also tried using a support vector machine for the same application and it performed slightly better.
hmmm.... no documentation, and the first demo I run fails due to an incompatibility with more recent version of MATLAB. Last update was 2002... I say this submission is dead.
I don't feel it is fully tested. So, you can't use it "off the shelf".
Hi, I am trying to load netlab into Matlab6 but the software does not recognise. Can someone help me....tks
could u please send a tutorial on netlab? i am trying to perform GMM on the probability density estimator rather than on samples (i.e., to break the pde i have to the creating gaussians).
hi,I have some problem with ant colony optimisation. can someone help me .thanks
I have used NetLab for for doing preliminary research and general benchmarking of Neural Network ideas. I find the toolbox very useful, especially when it is used with the accompanying text book.
Hi, I am trying to load netlab into Matlab ver 13 but the software does not recognise the commands used in netlab. Can someone help me....tks
Outstanding tool for use in addition to Matlabs nnet toolbox. Netlab incorporates concepts from baysien conditional type models into nnets. Reader particularly liked and found the parts from data sampling and optimization useful. Material bridges some of the gap between statistical data anaylsis and the control system approach used by the Matlab toolbox.
really helpful last download did not fully work
very helpful compilation of files
could u please send a tutorial on netlab
please send the tutorials on gaussian mixture model to my mail.
this site is very good