Stochastic resonance is an averaging method to amplify a data signal while allowing the noise to cancel out. It may be accomplished by adding a random white noise component to a stream that contains non random noise, or subtracting a non random noise if it can be characterized.

In general a random stream carries no information, and a non random stream always carries some type of information. If the transmission continues long enough or is repeated often enough to get a good data base the noise and information components can usually be separated.

This article seems serious enough, with an extensive formal reference section

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2660436/

Stochastic resonance process boost signals which are week by adding white noise.The frequency within the white noise which resembles that of the original signal will resonate with each other and will amplify the signal.

@Andre,

Thanks for the article, it seems quite relevant and well written. I would say that stochastic resonance (SR) is undefined because it is a syndrome of many different effects whose common characteristic is that authors do not understand them. This is a crucial ingredient: when authors come to understanding of what causes change in detector's response then the term "stochastic resonance" is not used anymore. That is why there is no definition of SR. In most effects named "stochastic resonance" randomness of the noise signal is NOT relevant (it could be a DC level or periodic signal) and in probably all cases resonance is NOT the mechanism for amplification. In most cases  simply changing the detection threshold or setting the operation point into a place where gain slope is steeper is what really happens. 

@Anirbid,

It is impossible that signal of a given frequency would constructively interfere with noise since the PHASE of any frequency in noise is fluctuating such that its interference with any single/frequency signal would average to zero. Also, when two frequencies constructively interfere  their amplitudes are just summed, there is no RESONANCE effect there! Furthermore if you believe that frequency from noise and signal would resonate, why would you not believe that  they would cancel each out? I believe in "stochastic cancellation" from now on! Any frequency in the signal will get cancelled by the same frequency but opposite phase in the noise. I think I just discovered a whole new mechanism for KILLING the signal!