Real-time artificial neural network regression

Example of artificial neural network (ANN) performing regression fitting in real-time. The bottom panel shows pulses of changing relative humidity (RH). A sensor material responds to the changing RH in the "response" panel. The ANN continuously learns the shape of the response and creates a model and prediction for the future response of the material, shown as "Model". The error between actual sensor response and the model is the residual, or percent error, in the top panel.