Most real world applications need to take account of costs, whether it is the cost of obtaining data, costs associated with applying a data mining algorithm or the financial and legal implications of misclassification. Thus, for example, in marketing, we are interested in identifying investments that produce greatest income given the cost of marketing. In medical diagnosis we need to take account of the costs of tests required in diagnosis,
such as blood tests, x-rays and scans.
This workshop includes papers that explore topics such as learning rules that take account of costs of misclassification, cost sensitive unsupervised learning, the use of utility and game theory in balancing costs and benefits, active learning for minimising costs of acquiring information and even how to take account of costs when they are missing. The workshop will provide a useful opportunity to understand some of the key issues in cost sensitive data mining and current direction of research in this field.
14:00:14:10 Welcome and Introduction, Sunil Vadera
14:10:14:40 Invited Talk: Cost Sensitive Action Rule Mining, Hendrik Blockeel
14:40-15:05 A Weighted SOM for classifying data with instance-varying importance, Peter Sarlin
15:05-15:30 When Additional Views Are Not Free: Active View Completion for Multi-View Semi-Supervised Learning,
Brian Quanz and Jun Huan
15:30-16:00 Coffee Break
16:00-16:25 A Multi-Armed Bandit Approach to Cost-Sensitive Decision Tree Learning
Susan Lomax, Sunil Vadera, Mohamad Saraee
16:25-16:50 Learning in the Class Imbalance Problem When Costs are Unknown for Errors and Rejects
Xiaowan Zhang and Baogang Hu
16:50-17:15 Learning Cost-Sensitive Rules for Non-Forced Classification, Arjun Bakshi and Raj Bhatnagar
17:15-17:40 Towards Utility Maximization in Regression, Rita P. Ribeiro
17:40-18:00 Discussion and Closing Remark