12:00, 22 Nov 1995, WeH 7220 IPC: Task Clustering and the Selective Transfer of Knowledge Across Multiple Learning Tasks: Throught and Results Sebastian Thrun Recently, there has been an increased interest in machine learning methods that learn from more than one learning task. Such methods have repeatedly found to outperform conventional, single-task learning algorithms when learning tasks are appropriately related. However, they usually fail to improve the results when tasks are not appropriately related. This talk investigates the feasibility of selectively transferring knowledge across multiple learning tasks. In particular, I will describe the task-clustering (TC) algorithm. TC clusters learning tasks into classes of mutually related tasks. When facing a new thing to learn, TC first determines the most related task cluster, then exploits information selectively from this task cluster only. An empirical study carried out in a mobile robot domain shows that TC outperforms its unselective counterpart in cases where many tasks are irrelevant.