In AI Input parameters is applied that results in an output response, This response is compared with target response, In Supervised Learning, a superviser helps in setting the synaptic weights so tht actual output matches with target output.
In contrast it does not require any superviser as there is no target output.Arbitrarliy organises different input patterns in to categories.Each new pattern is arranged in to a new category if nothing exists already.
It is similar to the Supervised Learning, but it does not help in setting weights for target output generation But It show PASS/FAIL indication.
In this case there are several agents at output layer which competes to produce output for input patterens.The one that produces output close to target wins, and others cease producing output for next same input pattern.It works on principle of the right person at right time at the right place.
AI Networks are characterized by
- Collective and synergistic computation
- Asynchronus in nature
Differnet Learning Algorithems
Adaline and MadLine modes
Back Propagation Algorithem
Adaptibe Resonance Theory
Memory Type Paradigms(RAM,CAM.TAM.BAM)
All this learning paradigms assume Clear Input patterns, When Input Patterns are fuzzy it needs an additional layerof complexity that defines variables and membership functions to create rules for input selection.
Subscribe to Octopus
Get the latest posts delivered right to your inbox