Long-term properties of time series generated by a perceptron with various transfer functions
A. Priel, I. Kanter
Abstract
We study the effect of various transfer functions on the properties of a time series generated by a continuous-valued feed-forward network in which the next input vector is determined from past output values. The parameter space for monotonic and non-monotonic transfer functions is analyzed in the unstable regions with the following main finding; non-monotonic functions can produce robust chaos whereas monotonic functions generate fragile chaos only. In the case of non-monotonic functions, the number of positive Lyapunov exponents increases as a function of one of the free parameters in the model, hence, high dimensional chaotic attractors can be generated. We extend the analysis to a combination of monotonic and non-monotonic functions.
Create a lesson
Related papers
Orbital cancellation and reentrant localization in a quasiperiodic trimer ladder
Xin Wang, Pei-Jie Chang, Quanfeng Lu et al.
Quantum Localization in Geometric Complex Networks
Jan Lewenstein-Sanpera, Tobias Grass, M. Àngeles Serrano et al.
Eigenstate transitions and critical transport in 3D Aubry-André model
Pragna Das, Lev Vidmar, Miroslav Hopjan
Weyl's theorem and the theoretical origin of Boson peak in two-dimensional scalar spring network
Cunyuan Jiang
On physical implementability of reinforcement-learning-driven reasoning
Eiji Konishi
The Stone-Wales trefoil: a new topological defect in graphene
M. V. Kondrin, Y. B. Lebed, V. V. Brazhkin