We assume the scenario that two tasks (S1 and S2) are given which were acquired with the same measurement equipment (e.g., two EEG channels measured with the same signal amplifier). The aim is to compute the correlation between these two tasks independently of the noise of the measurement equipment. To simulate this, we assume two sinusoidal signals are given (as shown in the upper plot, above), which are perfectly correlated. We add Gaussian noise N(0,f) to both tasks, where f = [0.01, 0.05, 0.1, 0.5, 1]. The first row of the above shows the original dataset (blue solid line) with f = 0.01; the second row of the above shows the data with f = 0.1, and the third row shows the data with f = 1. For each case, a MTGP model can be trained. The predictions are shown by the red solid lines and the confidence intervals by the grey shaded areas. As the noise on the signals is increased, the confidence intervals on the predictions increases as anticipated. However, it may be seen that the mean prediction results still describe the sinusoidal behaviour of both tasks.