FEATURE SELECTION FOR PATTERN RECOGNITION SYSTEMS

摘要: Publisher Summary The major problem of pattern recognition is essentially the discrimination input data between statistical populations via search for features among members a population. This chapter presents several approaches extraction in systems. design systems generally involves areas. first concerned with representation which can be measured from objects class. sensing problem. second selection characteristic or attributes received data. often referred to as feature third deals determination optimum decision procedures are needed process identification and classification. In solving problem, set parameters estimated optimized involved. gives rise parameter estimation has been recognized an important system. When complete discriminatory each class determined measurement, classification patterns will present no automatic may reduced simple matching procedure.