OpenNN - Open Neural Networks Library Reviews

OpenNN is an advanced open source class library which implements neural networks. This library has been developed in C++ programming language so that, it has better memory management and higher processing speed. It is really recommended. I have used it for different type of projects and it has generated successfully models.

Excellent C++ library for designing neural networks for different type of problems. It also has a large amount for preproccesing the data and testing the methods.

Great library to perform every desired action with neural networks. It has a good design and a perfect flexibility and generality to create the model that each problem requires. Nice new version solving some bugs and improving the documentation.

Just started using OpenNN. I like the design of the library, but implementation of the API leaves something to be desired, as there is complete and utter confusion over whether ownership of objects being passed in and returned passes with the object or remains behind. It has cost me a lot of time burrowing through the source to figure out what it is in each case, and hence fix up the crashes you get when you get it wrong. Why do people code C++ this way? Did they cut their teeth on Java or some other gaebage collected language where every object is owned by the system? I'm happy to fix the API, or at least document it to make it usable, but it seems the code published on github lags the released 3.0 version available on SF FRS. Is this some guys private code repository?

No usable documentation available, just a few examples and you need to understand on your own how to use it. There are some other libs around that are a lot better, at least for my use.

Excellent open source codes written in C++. A couple of bugs like the maxima and minima used during unscaling are wrong, due to which unscaling does not happen at the end. And similar smaller bugs. Overall great. I wish there was a support.

Nice work ! Great library.

It a shame it is not as popular as FANN. Terrific library. Found a bug with unscaling output that took me 11 hrs of work, yet still happy. I have reported it. Just in case: neural_network.cpp :line847: if(!unscaling_layer_flag) should be if(!unscaling_layer_pointer)