This web service provides an online demo of the Java classes for the AHN-algorithm. It introduces the simple AHN-algorithm that models a given dataset and returns the estimated values corresponding to the inputs.
If you are interested in the Java classes, please refer to the Resources sections in the previous page.
Click here to open the web service in a new window. Once you are in, follow the next steps:
1. Set the number of molecules. This is an integer value greater or equal to 2.
2. Set the learning rate. This is a real value between 0 and 1, both exclusive.
3. Select a CSV-file from your computer by using the Select File button. The rows of the file represent observations, the columns represent input features and the last column represent the target value. Only numerical values are allowed, and headers must be removed before using the CSV-file. See examples below.
4. Click on the Process Data button, and wait until the web service returns to you a CSV-file. Save it, if needed. The resultant CSV-file contains a single column representing the estimated values using artificial hydrocarbon networks.
5. If you want to see a comparative chart of your target values and the estimated values, click on the View Result Graph button. Click on the legends of the charts to hide/unhide them.
6. You can use this web service as long as you want.
NOTE: If the web server does not respond, wait for a minute and try again. It is possible that the server is not awaken.
Use the following prepared CSV-files that contain several useful examples:
Example 1 - Use AHN to model a piecewise function. Try with 3 or more molecules, and 0.1 of learning rate.
Example 2 - Use AHN to model a three-dimensional parametric function. Try with 5 or more molecules, and 0.01 of learning rate.