Premium
Active learning from process data
Author(s) -
Raju Gokaraju K.,
Cooney Charles L.
Publication year - 1998
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.690441009
Subject(s) - active learning (machine learning) , process (computing) , quality (philosophy) , computer science , domain (mathematical analysis) , machine learning , artificial intelligence , connectionism , artificial neural network , mathematics , mathematical analysis , philosophy , epistemology , operating system
Much of the prevailing connectionist machine learning research in chemical engineering assumes a one‐way passive relationship between the learner and the application domain. This article investigates a two‐way active relationship between learner and domain. An active relationship is useful and even necessary if the prevailing research is to be successfully applied to real‐world problems involving sparse and strongly biased data. A process development case study is used to illustrate the impact of data quality and quantity and to compare the performance of active learning against conventional passive learning. This study highlights on the need to assess data quality and demonstrates the improvements in the rate of active learning.