Learning by recording cases is one of the important topics in AI on learning. It is also related to Analogical reasoning. ###Recording Cases Memory contains a large of older problems with results. When a new problem is provided memory tries to relate with the closest similar problem for finding the results.
A simple example for doctors have all the patients symptoms so any new patients with similar symptoms come they will be able to easily identify the problem with the patient.
Similarly in programming we record all the errors which we have faced are recorded in memory. So when similar type error comes our memory will be able to retrieve the exact solution which fixed the error.
###Nearest Neighbour Method We use this method to find the most similar or most closest solution from the already available cases in memory. This solution might work when the cases or problem have two attributes which can be mapped in x and y axis. But when it grows this way of solving may not provide us the exact solution.
###K Nearest Neighbour method For various attributes we generalize the nearest neighbour method to be more generic to capture all the attributes. This will be really help full in real time problems.
- Problems that we face today are similar to problems we have faced earlier. Typical example would be tying a shoe lace.
- Learning and memory are very important since these two can equally compete against the reasoning part in the Cognitive architecture