1. Capable of self-organizing to an arbitrary sequence of sample patterns in stationary as well as non-stationary environments.
   
   
2. Ability to continuously learn new patterns without losing previously learned information.
   
   
3. Revelation of embedded rule set in the network can be accomplished in linguistic format that users can clearly comprehend.
   
   
4. Multiple Classifier Systems enable Pannasoft Ingenuity to produce better results for higher accuracy in decision making.

Pannasoft Ingenuity overcomes the limitation of Artificial Neural Networks

Neural networks have been empirically confirmed to be useful in many applications.  Nonetheless, most neural network models refused to tell what they have learned due to their distributed knowledge representation. Neural networks are regarded as a "black box" technology because of the lack of comprehensibility.  In general, comprehensibility is one of the required characteristics of reliable systems.  Therefore many works have been carried out to address the issue of improving the comprehensibility of neural networks. Rule extraction is a technique to overcome this problem.  The merits of including rule extraction technique as a supplement to neural networks are as follows:

1. Provide an explanation component to neural networks
2. Overcome the problem of knowledge acquisition
3. Explore data and induce scientific theories
4. Improve the generalization of neural networks solutions

Pannasoft Ingenuity with AAA is a hybrid neural networks based system, integrated with the rule extraction technique where it has revealed promising characteristic of building a genuinely autonomous system.  The revelation of embedded rule set in Pannasoft Ingenuity can be accomplished in the IF-THEN rule, which is straightforward and much more comprehensible.

<< back