| 12/29/2010 |
Neuroph 2.5 with Neuroph Studio beta Released |
The Neuroph 2.5b comes with exciting new features and improvements:
1. Neuroph Studio - new GUI based on NetBeans Platform!
2. Performance and algorithm optimizations
3. Integraton with Encog Engine
For full release announcement see http://neuroph.sourceforge.net/neuroph-2.5-released.html |
| |
|
| 10/02/2010 |
Neuroph at JavaOne materials available |
|
| See the videos, slides and pictures from Neuroph related sessions at JavaOne conference: http://neuroph.sourceforge.net/neuroph_at_java_one.html |
| |
|
|
| 09/25/2010 |
Encog switching to Apache license |
| Our partner project Encog http://www.heatonresearch.com/encog will switch to Apache license as of version 2.5. This will allow better license compatibility with Neuroph and other projects using Apache. Read official announcement by Encog team |
| |
|
|
| 09/20/2010 |
Neuroph on JavaOne |
Neuroph will be presented at two sessions at JavaOne 2010 conference (http://www.oracle.com/us/javaonedevelop/062264.html), the world's biggest and most influential conference about Java technologies.
Friends of the NetBeans Platform, Tuesday, 09.21.10. at 11:30AM
On this discussion panel we'll present new Neuroph IDE based on NetBeans Platform
BOF: Patterns for Modularity, Jaroslav Tulach, Zoran Sevarac, Anton Epple, Tuesday, 09.21.10. at 6:00PM
On this session we'll share our experience about porting Neuroph GUI to NetBeans Platform, and creating modular application |
| |
|
|
| 09/10/2010 |
Neuroph on Jax Enter |
| One of the Neuroph contributors, dr Valentin Steinhauern, wrote article in german and its published on Jax Enter http://it-republik.de/jaxenter/artikel/Neuronale-Netze-in-Java-3338.html well known german site about Java technology. |
| |
|
|
| 07/19/2010 |
Neuroph and Encog Collaboration Announcement |
| Two major open source neural network projects for the Java platform, Encog http://www.heatonresearch.com/encog and Neuroph http://neuroph.sourceforge.net , have announced collaboration on the development of advanced Java neural network technology. Read full article at http://neuroph.sourceforge.net/NeurophEncogCollaboration.html |
| |
|
|
| 07/09/2010 |
Improving Neuroph Performance |
| We've published an article with Neuroph performance analysis http://neuroph.sourceforge.net/improving_performance.html and released the development version of Neuroph 2.5a which contains backpropagation implementation optimized for high performance. Along with this we've included benchmarking code contributed by tahere as separate NetBeans project. We also announced development plans to improve Neuroph performance and we expect to get feedback from our community and contibutors. We allready managed to improve performance significantly, but we're going to try some more in order to be competitive with other frameworks. Also we have some results that may show that recent benchmarking might be misleading, but we need some more testing to do. |
| |
|
|
| 06/04/2010 |
Neuroph Benchmarking |
We opened Neuroph benchmarking section at our wiki: http://sourceforge.net/apps/trac/neuroph/wiki/TestResults
Currently there are three articles: one for Iris dataset and two for Neuroph, Encog and JOONE side by side comparasion.
The last two are very usefull for us to tell us where we are at the moment and what should we do in the future to be competitive with other Java neural network frameworks. Thanks to Tahere for these articles.
http://www.codeproject.com/KB/recipes/xor-encog-neuroph-joone.aspx
http://www.codeproject.com/KB/recipes/benchmark-neuroph-encog.aspx
The conclusion is that in future we have to provide high performance implementation based on matrices, support for multiple cores and maybe even using GPU. There is ongoing discussion about this at our forums https://sourceforge.net/projects/neuroph/forums/forum/862857/topic/3727747 |
| |
|
|
| 05/31/2010 |
Digit recognition with Neuroph on Android |
| Neuroph has successfully been used to create a digit recognition application on Android. Read the entire post at http://www.jdamcd.com/?p=84 |
| |
|
|
| 05/16/2010 |
Neuroph 2.4 released |
Version 2.4 of Neuroph has been released. The most important changes include:
- Backpropagation improvements (dynamic learning rate and momentum, min error stopping criteria, bias neurons)
- OCR tools and API
- Stock market prediction sample
- Perceptron and Multi Layer Perceptron learning vizualisation samples
- New methods to make API more comfortable to use
- Improved documentation
- License changed from LGPL 3 to Apache 2 licnese
- See CHANGELOG in released package for details.
This release brings significant improvements to backpropagation learning, overall performance and code quality.
Many thanks to all contributors and everyone else who gave usefull suggestions in discussions at our forums.
The 2.4 release is available for download here |
| |
|
|
| 05/09/2010 |
Final release 2.4 coming |
We're finilizing the release 2.4 and we'll make it available for downlod soon. Estimated time for release within two weeks.
The new release will bring the entire framework to the next level since it provides some important new features, learning algorithm improvements, tools, performance improvements, bugfixes and API extensions. Among them are the OCR tools and API. |
| |
|
|
04.29.10. Neuroph at Google Summer of Code 2010.
The project proposal titled "Implement a map/reduce enabled Neural Network with back propagation on Hadoop" which includes Neuroph has been accepted for the Google Summer of Code programme. Project is about making Neuroph to run on cluster computing framework Hadoop and enableing it to process and learn huge amounts of data. The project has been submited by student Zaid Md. Abdul Wahab Sheikh at Apache Software Foundation. Also this started the initiative to integrate Neuroph with Apache machine learning library Mahout. Read more about this at:
Original project proposal: https://issues.apache.org/jira/browse/MAHOUT-364
Discussion at Neuroph forums: http://sourceforge.net/projects/neuroph/forums/forum/862857/topic/3675039
Neuroph and Mahout integration: https://issues.apache.org/jira/browse/MAHOUT-383
Hadoop homepage: http://hadoop.apache.org
Mahout homepage: http://lucene.apache.org/mahout
Google Summer of Code 2010 Accepted Projects: http://socghop.appspot.com/gsoc/program/list_projects/google/gsoc2010
03.16.10. Neuroph on NetBeans Platform. Several Neuroph contributors went for the Certified NetBeans Platform Training at Faculty of Organizational Sciences. So now we're making official announcement that In the following months we'll be working on porting the existing Neuroph GUI easyNeurons to NetBeans Platform. Although it will take some of our development time to rebuild the existing features, when its done we'll have the first class GUI/IDE for neural network development and scalable platform for future development. At the same time we'll gain the NetBeans infrastructure for distributing, updateing and opportunity to provide integration with other well known applications based on NetBeans Platform.
03.04.10. Stock Market and Chicke Price Prediction Tutorials. Dr. Valentin Steinhauer has contributed two great tutorials about how to use Neuroph for stock market (or any other economics) prediction. Both tutorials come with full Java source and NetBeans projects. Take a look at stock market and chicken price tutorials.
01.20.10. Neuroph on NetBeans Zone. Third part of the interview titled Neuroph: Smart Java Apps with Neural Networks has been published on developers social network NetBeans Zone. Plans for migration of Neuroph's GUI to NetBeans platform in cooperation with NetBeans community announced!
12.24.09. NEAT support for Neuroph - preview release. We have published the preview release of NEAT support for Neuroph. NEAT stands for “Neural Evolution of Augmeted Topologies”, it is a Genetic Algorithm for the evolution of Artificial Neural Networks (ANN) developed by Ken Stanley.
The NEAT support for Neuroph is developed and contributed by Aidan Morgan.NEAT preview is available for download here
12.13.09. Time Series Prediction Tutorial by Laura E. Carter-Greaves. We have published tutorial for time series prediction and function approximation which shows the basics about how to do this with Neuroph. Send us your comments, and if you find it usefull we'll make some more. Read more
12.04.09. Neuroph 2.3.1 with Image Recognition released. This version brings important bugfixes for image recognition and several GUI improvements. This version fixes the following bugs:
- fixed issue with editing gui in NetBeans (fixed NetBeans project file)
- fixed LMS formula
- fixed testing in black and white mode for image recognition
- fixed gui bug - exceptions when creating large networks
- changed image recognition API, so the color mode is automaticaly detected from settings used for network training removed unnecessary methods
- graph view - migrated graph view to JUNG to 2.0, created specific network layouts and removed unnecessary options
-
ANT build file is now included in release which can build the jars for library and GUI.
» Download this version here
12.01.09. 3D Neural Network Visualization. The Neuroph and SpaceNet (http://automenta.com/spacenet) projects are collaborating on a new kind of 3D environment for visualizing neural systems.
Developers of the project share a vision of an intuitive tool for evolving and exploring the possibilities of neural networks, so future work may also include environment for 3D editing and interaction with neural systems. The developers are excited about the new features being designed.
The common software component responsible for the (easy) integration is the JUNG Graph API (http://jung.sourceforge.net).
Demo video available at: http://www.youtube.com/watch?v=c7cGSb0ITQ4
10.09.09. Neuroph 2.3 with Image Recognition released. This version brings neural network image recognition support with easy to use Java library and GUI tool. Image recognition can be done in just one method call! In addition this release also comes with improved API, better javadoc documentation and some basic samples.
08.19.09. Neuroph on NetBeans Zone. Two-part interview titled Neuroph: Smart Java Apps with Neural Networks has been published on developers social network NetBeans Zone. Take a look at part 1 and part 2 and and get to know the Neuroph project behind the scene, as well as our future plans. Part two of the interview also contains very brief howto for the Neuroph GUI.
07.07.09. Neuroph on Gentoo Linux. The latest version of Neuroph Framework has been included in Gentoo Linux package repository. The GUI application easyNeurons is incuded as a separate package. This is a great honour for our project and we're sure this will help to grow the Neuroph community. Thanks to Alistair Bush for this.
Read more about Gentoo Linux here |
 |
07.05.09. Neuroph v ersion 2.2 has been released. This version brings important improvments and some usefull new features:
- Momentum for backpropagation
- Import training set from txt files
- Network error graph during the training
- Support for Instar, Outstar and BAM networks
- XML support for neural networks and training sets
- Basic help system
- Sample image recognition application
05.14.09. Neuroph on Slashdot. An article about how Neuroph has been used in DotA Script has been published at Slashdot! This is a big promotion for Neuroph.
05.14.09. New version of DotA Script released. The new version this game script uses the image recognition library, and the newest learning algorithms which will be available in the next release of Neuroph. Read here more about how Neuroph neural networks are being used for image recognition.
04.07.09. Version 2.1.1.beta has been released. This version brings few bug fixes for previous release and some new features:
- kohonen network sample had bug with drawing - fixed
- backpropagation with tanh was throwing exception - fixed
- added basic neuron sample
- added unsupervised hebbian learning
- added oja learning rule
- added new GUI components JNeuron, JLayer and JNeuralNetwork which provide cleaner API and new features
like neuron color depending of its activation level
03.31.09. Version 2.1.0.beta has been released! This version brings the following improvments:
- completly separated core framework library neuroph.jar and gui application easyNeurons.jar
- cleaned up most of the compilation warnings
- removed deprecated stop() methods for stopping learning thread, and restructered learning classes for IterativeLearning
- added foreach loops wherever applicable, which significantly improved code readability
- improved serialization support
- several bug fixes in core and GUI
- neural network plugins support (see plugins package)
- labels plugin which enables labeling for all neural network componets
- removed borland layout managers in GUI
- documentation for all classes and methods from library
This is now recommended and most stable release.
03.17.09. Web site updated in many sections. Some of them which are important:
03.07.09. Version 2.0.0.alpha has been released publicly. This version comes with new GUI application, support for new network architectures, learning rules, including competitive networks and Hebbian learning, cleaned code, and better documentation. Please note that this version is not compatible with previous versions, but the users of previous versions can resolve compatibility issues easily. This is now the recomended release.
03.07.09. Version 1.0.1.alpha has been released. This version adds serialization support to version 1.0.0 and various optimizations thaks to Jon Tait.
03.06.09. Neuroph has been successfully applied for image recognition in game! See http://dota-autoscript.com
|
