Analysis Of Q- Learning Based Game Playing Agents For Abstract Board Games With Increasing State-Space Complexity