A partner of the TRIZ consulting company Ideation International, Inc., Michael, from Taiwan, sent us this remark/question:
«According to my knowledge, the NN should have three functions: classification, prediction, and optimization. Hope I am right.»
Michael
— Yes, you’re right, but this definition is too narrow and very limited. Actually, PANN provides much more. Let me elaborate.
The initial function of thinking, both in humans and machines, is the recognition of any images, visual, verbal, odor, tactile, etc. The thing is that any image (portrait, table, graph, formula, and so on) contains many different patterns. Any recognition is carried out precisely through patterns. A neural network is just a filter that highlights strong patterns in a processed image, by which the memory can understand (recognize) what it is. In human thinking, recognition is equivalent to the operation of abstraction. And all psychological operations in the human head, one way or another, boil down to abstraction:
- Making choices
- Classification and clustering
- Approximation and extrapolation
- Optimization
- Processing and transformation of images
- Modeling of any processes
- Information encoding and compression
- Prediction of functioning and development
- Management of any systems
- System-creative processes
Thanks to its high recognition ability, PANN makes it possible to implement all this, and thus actually create Artificial Intelligence.
The main advantage of PANN is fast and reliable training. But also very important is the high controllability of both learning processes and recognition processes and the use of recognition results, which is completely absent in classical networks. Therefore, the PANN network can work very well in human-machine usage mode.
Another important feature of the PANN network is that it is very well compatible with our software in literally all areas of our work:
- Efficient search for information
- Solving inventive problems for improving and optimizing systems (IPS) in both engineering and science and non-technical areas – logistics, business, social issues
- Development forecasting and management (DE)
- Failure analysis and failure prediction
- IP evaluation and HIP enhancement
- Creative education, etc.
Answered by Boris Zlotin, TRIZ specialist, chief scientist, CTO at Omega Server, Inc.