Chaos Engineering and Its Application to Parallel Distributed Processing With Chaotic Neural Networks

Deterministic chaos has an intriguing characteristic that definite deterministic dynamics produces erratic and long-term unpredictable behavior [1], [2]. Understanding of deterministic chaos has greatly progressed in the past 25 years, although the possibility of chaotic dynamics was demonstrated mathematically at least 100 years ago by Hadamard and Poincaré [3]. It is now well known that chaotic phenomena exist ubiquitously in both real-world nonlinear systems and mathematical models. Chaotic dynamical systems have many interesting properties, such as nonperiodic and complex temporal behavior, sensitive dependence on initial conditions or the so-called butterfly effect, fractal structure, and long-term unpredictability. Moreover, chaos has many potentially useful functions. For example, Conrad discussed that chaos can be used, with respect to adaptability theory and complex biological systems, for such functions as search with