Stochastic Resonance in Neural Network, Noise Color Effects

Some systems cannot be predicted by classical theories and it is required the development of combined deterministic and stochastic theories that make used of noise for dynamical prediction. Noise is not always an interfering signal which perturbs the system. On the contrary, noise signal can enhance the performance of some non-linear systems. The advantage of noise is observed in Stochastic Resonance (SR) where noise is used for amplification and subsequent detection of small signals. To detect this phenomena it is necessary that the system have a bistable potential barrier that creates a threshold, the input of the system should be a weak periodic signal which amplitude is below threshold together with an stochastic signal. In this way, the SR is detected when there are weak periodic signals that are added to different noise colors in order to be amplified and optimised. The interactions between the two signals transform the potential of the system precisely at the frequency of the weak periodic signal that is added to the system. The behaviour of the SR is detected in a neural network and it is study under noise color variations. Here I found that Pink noise amplified the sub-threshold input signal twenty times more in comparison to white noise. This is evidence of the functionality of background noise in the brain, where neurons are naturally embedded in pink noise.