Bootstrapping training-data representations for inductive learning: a case study in molecular biology

摘要: This paper describes a "bootstrapping" approach to the engineering of appropriate training-data representations for inductive learning. The central idea is begin with an initial set human-created features and then generate additional that have syntactic forms are similar human-engineered features. More specifically, we describe two-stage process good learning: first, generating by hand (usually in consultation domain experts) seem help learning, second, off these developing applying operators new look syntactically like expert-based Our experiments DNA sequence identification show successful representation data can be expanded this fashion yield dramatically improved results