I don't think that AlphaGo was trained more on Lee Sedol's games, than on others' games. The team said that they can't find computer weaknesses until AlphaGo plays against top caliber.