A Rigorous Verification of Chaos in A 3D Neural Network
Fangyan Yang, Ping Zhou, Qingdu Li · 2006
This paper proposes a rigorous verification of chaos in an artificial neural network (ANN) consisting of three neurons by virtue of the topological horseshoe theory. Because the two disjoint compact subsets producing a horseshoe we found in a certain cross-section both expand in one directions under the forth Poincare return map, we conclude that there exists chaos in the ANN