Feb 28, 2012
Stochastic-Based Pattern Recognition Analysis is a paper that begins to exploit the many advantages of stochastic computing by applying it to the process of pattern recognition. The problem is to compare an input signal with a reference signal to determine whether they match.
In a nutshell, stochastic computations outperform the conventional techniques in the following ways:
They offer simple multiplication with AND gates comparable to ease of addition.
Stochastic computers are robust to noise.
Ability to process signals in parallel, making it faster than serial computations.