Robotic rehabilitation of the hands from a neuromuscular impairment such as stroke requires controllers that could provide subject-specific assistance and result in fastest possible recovery. We present two such assist-as-needed controllers for a hand exoskeleton called Maestro that is designed to provide accurate torque assistance to a subject. Learned force-field control is a novel control technique in which a neural-network-based model of the required torques is learned offline for a specific subject and then used to render a force-field to assist the finger motion to follow a target trajectory. Adaptive assist-as-needed control, on the other hand, estimates the coupled finger-exoskeleton system torque requirement of a subject using a radial basis function (RBF) network and adapts the RBF magnitudes in real-time to provide a feedforward assistance for accurate trajectory tracking. Experiments with a healthy subject on Maestro showed that while the force-field control is nonadaptive and there is less control on the speed of execution of the task, it is safer as it does not apply increased torques if the finger motion is restricted. On the other hand, adaptive assist-as-needed controller adapts to the changing needs of the coupled finger-exoskeleton system and helps in performing the task with a consistent speed, however, applies increased torques in case of restricted motion and is therefore, potentially less safe.