11:00 - 11:30
Rules based systems were the traditional way to build expert systems and to automate decisions, essentially domain knowledge coded into a table as predicates with precedence and scored by a rules or inference engine or it was Domain knowledge coded by programmers in Java, C++, SQL, Perl etc. and scored in the code.
Then came the Machine Learning Approach, where the system would learn from historical data with known outcomes using learning algorithms, Build models during training phase, For unknown input, use the prediction algorithm to evaluate output and eventually Rules are automatically ‘found’ by the algorithm that maps inputs to outputs.
Now, we have Shallow and Deep learning which has Multiple layers in neural network with intermediate data representations to facilitate dimensional reduction, Interpret non-linear relationships in the data and Derive patterns from data with very high dimensionality
The panel will discuss
Approaches to analytics and when to apply which or a combination of the above (Use cases)
What are the drivers for selecting an approach