作者: Prem Melville , Foster Provost , Maytal Saar-Tsechansky , Raymond Mooney , None
关键词: Expected utility hypothesis 、 Training set 、 Machine learning 、 Artificial intelligence 、 Estimation 、 Computer science 、 Value (computer science) 、 Feature (computer vision) 、 Missing data 、 Active learning (machine learning) 、 Data mining 、 Unit cost
摘要: … However, when many inexpensive features are also uninformative, Cheapest-first performs worse than a random acquisition policy (Figure 2(b)). Sampled Expected Utility, however, …