Enhancement of Closed-Loop Cognitive Stress Regulation Using Supervised Control Architectures.

We propose novel supervised control architectures to regulate the cognitive stress state and close the loop. We take information present in underlying neural impulses of skin conductance signals and employ model-based control techniques to close the loop in a state-space framework. For performance enhancement, we establish a supervised knowledge-based layer to update control system in real time. In the supervised architecture, the controller parameters are being updated in real-time. Statistical analyses demonstrate the efficiency of supervised control architectures in improving the closed-loop results while maintaining stress levels within a desired range with more optimized control efforts. The model-based approaches would guarantee the control system-perspective criteria such as stability and optimality, and the proposed supervised knowledge-based layer would further enhance their efficiency. Outcomes in this in silico study verify the proficiency of the proposed supervised architectures to be implemented in the real world.