You do not have permission to edit this page, for the following reason:
The action you have requested is limited to users in the group: Users.
Analysis & Visualisation of data Providing feedback for supporting instructors Predicting student performance Recommendations for students Student modelling Grouping students Social Network Analysis (SNA) Developing concept maps Constructing courseware Planning/scheduling/monitoring Experimentation/observation
Once registered form, specified format(s) will appear in the selection list above and will become available for every users Add format(s):
Web crawler Screen scraping Flat file database/Logfile extractor Structured database extractor
Simple data format conversion Simple data transformation operations Advanced data transformation operations Mathematical transformation of data for analysis
Basic statistics and data summarization Data mining methods and algorithms
Sequential Graphic Slide Show Chart/Diagram Map Tag Cloud Model Drawing Video/Animation Photograph Poster Game
Visualisation is interactive Visualisation is updated in "real time"
Students/Learners/Consumers Teachers/Tutors/Managers Developers/Designers Researchers Organisations/Institutions/Firms Others
= Description =
'''Neural Designer''' is a desktop application for data mining which uses neural networks, a main paradigm of machine learning.
Neural Designer follows an open core model by using OpenNN inside its learning engine. OpenNN is a popular open neural networks library written in the C++ programming language. It is hosted at SourceForge and licensed under the GNU Lesser General Public License.
The program supports both Windows and Linux platforms.
Neural Designer has two main windows, plus a calculation engine running in background. These allow you to (1) edit your settings, (2) perform computations, and (3) see your output. The next figure illustrates the flow of information in the software. The input here is a data set, and the output is a neural model.
[[File:Neural Designer activity.png|800px|thumbnail|center]]
The main components of Neural Designer are explained in detail below.
== Neural Editor ==
The editor lets you see and manipulate your settings. The next picture shows the main components of Neural Editor.
[[File:Neural Designer tasks.png|650px|thumb|center]]
As we can see, Neural Editor has three panes:
# A data book with data set, neural network, performance functional and training strategy tabs. It contains the settings for all the components needed to solve a given application.
# A task manager with data set, neural network, performance functional, training strategy and testing analysis tasks. It contains a list of task for each component to be run by the engine
# An output pad for the editor, the engine and the viewer. Most information, warning and error messages will appear here.
== Neural Engine ==
The engine runs tasks. That component does not have any window, but it executes processes in background. Tasks are called from Neural Editor. The results will show up in Neural Viewer.
As we have said, the engine is developed using the high performance neural networks library OpenNN.
== Neural Viewer ==
The viewer writes a report displaying comprehensive and visual results from tasks. The following figure illustrates the Neural Viewer window.
[[File:Neural Designer histograms.png|650px|thumbnail|center]]
= Learning tasks =
Two of the most important learning tasks that Neural Designer solves are function regression and pattern recognition. Both of them use data sets to construct the predictive model
== Function regression ==
The objective here is to design a model which makes good predictions for new data, in other words, one which exhibits good generalization. This learning task is illustrated here through an example. In particular, we will model the performance of an engine.
The first step is to prepare the data set, which is the source of information for the function regression problem. Neural Designer contains different utilities for alerting on the presence of spurious data, detecting outliers, etc.
The neural network defines the model mentioned before. It must be trained in order to learn the underlying relationships between the outputs and the inputs. The next picture illustrates the neural network for this example.
[[File:Neural Designer engine network.png|550px|thumbnail|center]]
The neural network must be tested against data that it has never seen. A useful method for that is to perform a regression analysis between the outputs from the neural network and the corresponding targets in the testing data.
The next picture shows the results of this testing method.
[[File:Neural Designer engine plot.png|600px|thumbnail|center]]
The statistics on the error data measure the minimums, maximums, means and standard deviations of the errors between the neural network and the testing instances in the data set. The next tables show the error data for the two targets in this example.
[[File:Neural Designer data errors.png|450px|thumbnail|center]]
The last step is the production phase. Once the model has been tested, we can use it to predict the output. For that, we can use the explicit expression reported by Neural Designer, as illustrated below.
== Pattern Recognition ==
In this learning task, the aim is to design a neural network that can predict the correct class for given attributes. The central goal here is to design a model which makes good classifications for new data.
The first step is to prepare the data set. The data for this example has been taken from a breast cancer diagnosis application.
The next table depicts the basic statistics (minimums, maximums, means and standard deviations) of all variables.
[[File:Neural Designer data statistics pattern.png|550px|thumbnail|center]]
The second step is to represent the classification function. Neural Designer includes innovative neural network models for solving pattern recognition applications. A graphical representation of the network architecture in our example is depicted next.
[[File:Neural Designer cancer network.png|600px|thumbnail|center]]
The next step is to test the generalization performance of the trained neural network. Here we compare the values provided by this technique to the actually observed values. The next table shows the confusion matrix, which depicts the number of instances correctly and incorrectly classified.
[[File:Neural Designer confusion matrix.png|500px|thumbnail|center]]
The following table depicts the binary classification performance parameters for the trained neural network on the testing instances.
[[File:Neural Designer binary classification.png|300px|thumbnail|center]]
The neural network is now ready to predict outputs for inputs that it has never seen.
This is a minor edit
Watch this page