Abstract:In this paper, the recurrent neural networks (RNNs) with a generalized activation function class is proposed. In this proposed model, every component of the neuron's activation function belongs to a convex hull which is bounded by two odd symmetric piecewise linear functions that are convex or concave over the real space. All of the convex hulls are composed of generalized activation function classes. The novel activation function class is not only with a more flexible and more specific description of the activation functions than other function classes but it also generalizes some traditional activation function classes. The absolute exponential stability (AEST) of the RNN with a generalized activation function class is studied through three steps. The first step is to demonstrate the global exponential stability (GES) of the equilibrium point of original RNN with a generalized activation function being equivalent to that of RNN under all vertex functions of convex hull. The second step transforms the RNN under every vertex activation function into neural networks under an array of saturated linear activation functions. Because the GES of the equilibrium point of three systems are equivalent, the next stability analysis focuses on the GES of the equilibrium point of RNN system under an array of saturated linear activation functions. The last step is to study both the existence of equilibrium point and the GES of the RNN under saturated linear activation functions using the theory of M-matrix. In the end, a two-neuron RNN with a generalized activation function is constructed to show the effectiveness of our results.

Delayed Recurrent Neural Networks with Global Lipschitz Activation Functions

On the Global Stability of Delayed Neural Networks

Global Exponential Stability of Recurrent Neural Networks with Time-Varying Delay

Global asymptotic stability for a class of neural networks with time-varying delays

Global Robust Stability of Delayed Neural Networks with A Class of General Activation Functions

Improved Delay-Dependent Stability Condition of Discrete Recurrent Neural Networks with Time-Varying Delays

Deep Neural Networks with Trainable Activations and Controlled Lipschitz Constant

Global Convergence of Delayed Neural Network Systems.

Absolute Exponential Stability of Recurrent Neural Networks With Generalized Activation Function

Novel Global Exponential Stability Criterion for Recurrent Neural Networks with Time-Varying Delay

An estimation of the domain of attraction for recurrent neural networks with time-varying delays

On Global Exponential Stability Of Cellular Neural Networks With Lipschitz-Continuous Activation Function And Variable Delays

New Approach for Robust Stability Analysis of Delayed Recurrent Neural Networks

Graph-Adaptive Activation Functions for Graph Neural Networks

A New Artificial Neuron Proposal with Trainable Simultaneous Local and Global Activation Function

Multistability analysis of delayed recurrent neural networks with a class of piecewise nonlinear activation functions

Latent Assistance Networks: Rediscovering Hyperbolic Tangents in RL

Stability Analysis for Delayed Neural Networks via a Generalized Reciprocally Convex Inequality

Generalized Lmi-Based Approach to the Global Asymptotic Stability of Discrete-Time Delayed Recurrent Neural Networks

Global Exponential Stability for Complex-Valued Recurrent Neural Networks With Asynchronous Time Delays

Global Asymptotic Stability Analysis of Neural Networks with Time-Varying Delays