N++2 is a multi-threaded neural network simulator for simulating large
neural networks with millions of weights. During the training procedure,
it makes heavy use of multi-core CPUs using several threads for propagating
multiple training patterns in parallel. Relying on CBLAS calls where 
appropriate, n++2 is also able to benefit of a CPU's SIMD-capabilities.

N++2 is able to speed-up training of "simple" multi-layer perceptrons with 
only one or two hidden layers and a few hundred connections. But its main
purpose is to simulate huge neural networks with a number of hidden layers
and up to millions of weights in a "deep learning" setting. For this 
purpose, n++2 offers additional facilities for constructing symmetric 
autoencoder neural networks, easy layer-wise pre-training and for arranging
neurons in two-dimensional layers (helpful for processing images). Besides
fully connected layer types n++2 also comes with an implementation of sparse
connection structures, including receptive fields and shared-weights. Thus,
n++2 can also be used to simulate LeCun's convolutional neural networks and
to combine these sparse techniques with deep learning and layer-wise 
pretraining. 

N++2 uses Martin Riedmiller's Resilient Propagation (RProp) for fast and 
reliable training of the neural networks. N++2 builds on Riedmiller's 
original (non-parallel) implementation n++ and uses mainly the same interface 
(API) to the "simulator core", although the internal structure of the neural 
networks in n++2 is different. On a dual quad-core CPU with two threads
per core (16 threads in parallel) n++2 is about 27 times faster than the
original n++.

The latest (development) version of N++2 can be downloaded from GitHub:
git clone https://github.com/salange/NPP2
N++2 comes under the BSD licence. Feel free to use, modify and redistribute.
For more information please visit us at our site 
http://ml.informatik.uni-freiburg.de
or contact us directly via Email: 
Machine Learning Lab at University Freiburg: ml@informatik.uni-freiburg.de 
Original developer and present maintainer: sascha77@googlemail.com

The further development of n++2 will be an ongoing process open to the 
public. We're open for any contributions and appreciate any help in improving
the code, build-system and documentation. If interested to participate, 
please don't hesitate to contact us!


This directory includes the following subdirectories:

src             source code for n++2
doc             documentation (latex and doxygen-generated files)
demo_src        demo source code of applications
examples        network descriptions and pattern sets
npp2.xcodeproj  project file (Xcode 4.0)

The following subdirectories will be created during the build procedure:

include         headerfiles to be included from application programs
lib             object files to be linked to application programs
build           build files (Xcode)
bin             executable files


N++2 was tested on:
- OS X Lion   (Xcode and cmake)
- Ubuntu 10.04 LTS  (cmake)


Building n++2 using cmake:

Create a build-directory for storing configuration and temporary files:
> mkdir build_cmake

Change to directory:
> cd build_cmake

Configure cmake and create make files (also builds the demos):
> cmake -DDEMOS=ON ..

Build and install n++2:
> make
> make install


Create API-documentation from sources:
> make doc

The documentation can be found in
doc/api/html/