Noisy-Output-Based Direct Learning Tracking Control With Markov Nonuniform Trial Lengths Using Adaptive Gains | Synapse