A framework for generating data to simulate changing environments

摘要: A fundamental assumption often made in supervised classification is that the problem static, i.e. description of classes does not change with time. However many practical tasks involve changing environments. Thus designing and testing classifiers for environments are increasing interest importance. number benchmark data sets available static tasks. For example, UCI machine learning repository extensively used by researchers to compare algorithms across various domains. No such datasets Also, while generating relatively straightforward, this so The reason an infinite amount changes can be simulated, it difficult define which ones will realistic hence useful. In paper we propose a general framework simulate gives illustrations how encompasses types observed real also two most popular simulation models (STAGGER moving hyperplane) represented within.