| Automated discovery is a subfield of Machine Learning in the field of Artificial Intelligence which attempts to discern concepts or classifications from merely examining examples or observations. As with other AI programs, discovery programs often involve intensive utilization of computer resources. Discovery programs that run faster or are able to analyze larger sets of data would be helpful to other researchers desiring to utilize discovery programs in their own work.;This research explores ways to improve the performance of discovery systems by enabling their execution on parallel architectures. Two approaches to parallelizing these programs are offered. Potential obstacles to parallelization of discovery programs are discussed. Finally, results of successful parallelizations of two well known discovery programs, AutoClass and SUBDUE, are presented. |
