scientific article; zbMATH DE number 1047938
We consider a problem of defining a logic network as a higher-order neural network which has a given (arbitrary) state function for its Lyapunov function. The Lyapunov functions treated here are strict in the sense that a lower function value is realized if the corresponding state transition is permitted by an operation mode under consideration. In this note, we first review our necessary and sufficient condition for the existence of state functions, and clarify their relations with the derivatives of state functions. Considering an asynchronous operation mode, we can define local transition functions in such a way that the given state function is really a strict Lyapunov function for the network. The result is compared with the steepest descent method of function minimization problem for continuous variables. We also apply this result to higher-order positive term representation of state functions [the author, Inf. Process. Lett. 41, No. 2, 61-66 (1992; Zbl 0743.68132)]. Then a generalized quasi-symmetric condition is shown to be necessary and sufficient for the existence of a strict Lyapunov function. For the entire collection see [Zbl 0868.00046].
- Macroscopic Lyapunov functions for separable stochastic neural networks with detailed balance.
- Low-rank kernel approximation of Lyapunov functions using neural networks
- Lyapunov Functions for Neural Nets with Nondifferentiable Input-Output Characteristics
- scientific article; zbMATH DE number 32622 (Why is no real title available?)
- scientific article; zbMATH DE number 1047893 (Why is no real title available?)
- scientific article; zbMATH DE number 1972774 (Why is no real title available?)
- Efficient Construction of Neural Networks Lyapunov Functions with Domain of Attraction Maximization
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4348019)
