英语翻译In the real world concepts are often not stable but chan

英语翻译
In the real world concepts are often not stable but change with time.Typical examples
of this are weather prediction rules and customers’ preferences.The underlying data
distribution may change as well.Often these changes make the model built on old data
inconsistent with the new data,and regular updating of the model is necessary.This
problem,known as concept drift,complicates the task of learning a model from data
and requires special approaches,different from commonly used techniques,which
treat arriving instances as equally important contributors to the final concept.This
paper considers different types of concept drift,peculiarities of the problem,and gives
a critical review of existing approaches to the problem.