This document discusses using machine learning algorithms for job scheduling in a grid computing environment. It aims to minimize makespan, the total time to complete all tasks, by learning from past scheduling experiences. It proposes using ant colony optimization, where artificial ants probabilistically choose task-machine pairs to incrementally find optimal schedules. The algorithm is compared to other scheduling methods and extended to online scheduling by classifying jobs with attributes to appropriate machines. A feasibility study demonstrates classification and scheduling of test jobs using machine learning tools.