To some extent, I personally consider strength a moot point. All of these locks, if implemented right, easily have the strength to handle anything I would throw at it. However, theoretically, I'd say the axis, rolling lock, arc lock, and integral lock are all probably a step above the liner lock and lockback in strength.
Reliability depends on both the basic strengths of the design, and the actual execution. So far, the lock I'm most confident of is the axis. The integral lock is also very well-proven in my mind. On paper, the integral lock looks the simplest and hence the most reliable, but I feel that the vulnerability of tang-block locks to torquing doesn't give it any practical advantages over its more-complex cousins. The arc lock has tons of potential, but just hasn't been out quite long enough for me to be completely sure of it.
Although I agree with the sentiment above that I wouldn't buy a knife solely on the lock, the reverse isn't true for me -- I do avoid a knife based solely on the lock at times. I personally don't have enough faith in the reliability of liner locks to buy one. I feel the lockback is a well-proven lock (although there are some known vulnerabilities to test for), especially at lower price points, although some very beefy and reliable lockbacks at mid-end and even high-end price points are also around, e.g., Spyderco's Chinook on the production end, and Steve Mullin's incredible utility folder on the semi-production end. At the high end, if a knife had an axis, integral, rolling lock, or arc lock, and I otherwise liked the knife, I don't think I'd hesitate in buying it.