CMU is probably the best place for ML. They have their own Machine Learning department: http://www.ml.cmu.edu/ CMU is, IMO, only rivaled by Berkeley (since they have Michael Jordan, who is to ML as the other Michael Jordan is to basketball) and MIT with their AI Laboratory.
UT Austin would probably be second on your list. Cornell, Columbia, and UCSD are, obviously, all great schools, but I think the only way to compare them would be to specify an ML subfield. U. Washington, Standford, and Georgia Tech all have outstanding ML research as well and are probably better overall in ML than the above three. In fact, look for UW to climb towards the top in the coming years; they recently poached some big names (Ben Taskar, Carlos Guestrin, and Emily Fox) from CMU and Penn. Lastly, don't overlook UMass Amherst. Although they are not generally regarded as an 'elite' CS department, their AI/ML is very strong and ranks higher than Cornell, Columbia, and UCSD in US News' AI rankings.