Multistability of switched neural networks with sigmoidal activation functions under state-dependent switching☆
Introduction
As known widely nowadays, neural networks play important roles in many technical areas, such as pattern recognition (Kwan and Cai, 1994, Suganthan et al., 1995, Zeng et al., 2005), associative memory (Isokawa et al., 2008, Zeng and Wang, 2008, Zeng and Wang, 2009), and other areas (Bao et al., 2018a, Bao et al., 2016). In recent decades, various neural network models, including Hopfield-type neural network model (Huang and Cao, 2010, Zeng et al., 2016), cellular neural networks with memristors (Bao and Zeng, 2013, Di et al., 2017, Duan et al., 2015), Cohen–Grossberg neural network model (Gang et al., 2007, Wang et al., 2006, Wang and Zou, 2002, Zhang and Wang, 2008), switched neural network model (Bao et al., 2018b, Li and Cao, 2007, Li et al., 2009, Liu et al., 2013, Zhao et al., 2016, Zhao and Zhao, 2017), have been developed and analyzed.
Because of the complexity of actual environment, the connections of neurons in a network may change frequently. Due to the link failure or a new creation of the connection topology in the network, the switching between some different connection topologies are inevitable (Daafouz et al., 2002, Zhao et al., 2009). To describe this switching phenomenon in the network, switched system is introduced. A switched system, consisting of a number of subsystems or configurations modeled by using differential or difference equations, operates in a mode switching among these subsystems. Generally speaking, two categories of switching modes are usually considered (Liberzon and Morse, 2015, Persis et al., 2003). One is time-dependent switching, i.e. the switching rule is controlled only by time. For example, a continuous switched system is characterized by the following differential equation: Let , the parameter takes the value in the index set , each map in is assumed to be locally Lipschitz, and is the switching signal. The other type is state-dependent switching. In this case, the switching signal depends only on state . Compared with the switched systems under time-dependent switching, switched systems under state-dependent switching with different initial values may take on different equations at the same instant. The dynamical analysis of switched systems is much more complicated and challenging than that of conventional ones. Due to the significant values of switched systems in both theory and practice, stability analysis of switched systems is an attractive topic; e.g., Li et al., 2005, Wu et al., 2011, Yuan et al., 2006 and Zhang, Tang, Miao, and Du (2013).
Multistability is a notion to characterize the coexistence of multiple stable equilibrium points or periodic solutions (Cheng et al., 2006, Liu et al., 2016a, Liu et al., 2016b, Liu et al., 2017a, Liu et al., 2017b, Liu et al., 2017c, Liu et al., 2018, Nie and Zheng, 2015, Wang, 2014, Wang and Chen, 2012, Wang et al., 2009, Wang et al., 2010). It is well recognized that multistability analysis of neural networks depends critically upon the type of activation functions. With different activation functions, different multistability stability criteria can be derived. For example, in an -neuron Hopfield neural network with Mexican-hat-type activation functions, there exist at most equilibrium points and at most of them are locally stable and others are unstable (Wang, 2014, Wang and Chen, 2012). In a neural system with Gaussian activation functions, there exist equilibrium points, in which are exponentially stable, where (Liu et al., 2017a). In a neural system with piecewise-linear nondecreasing activation functions, there exist equilibria, in which are locally exponentially stable, where is the number of pieces in the piecewise-linear activation functions (Wang et al., 2010). In a neural network with discontinuous Mexican-hat-type activation functions under reasonable conditions, there exist equilibrium points in and of them are located at the continuous part of the activation functions (Nie & Zheng, 2015).
Numerous results are available for the stability analysis of the switched systems, e.g., Branicky, 1998, Chen et al., 2010, Guo et al., 2018, Guo et al., 2019, Guo et al., 2014, Hu et al., 2015, Kahloul and Sakly, 2018, Li and Cao, 2007, Lian and Wang, 2015, Long and Wei, 2007, Long et al., 2013, Nie and Cao, 2015, Niu et al., 2017, Singh and Sukavanam, 2012, Song et al., 2018, Song et al., 2006, Tang and Zhao, 2017 and Yu, Fei, Yu, and Fei (2011). The global stability of switched neural networks and the control of switched systems are addressed in Branicky, 1998, Chen et al., 2010, Hu et al., 2015, Kahloul and Sakly, 2018, Li and Cao, 2007, Long and Wei, 2007, Long et al., 2013, Niu et al., 2017, Singh and Sukavanam, 2012, Song et al., 2018, Song et al., 2006, Tang and Zhao, 2017 and Yu et al. (2011).
In contrast to global stability, few results are available on the multistability of switched neural networks, especially under state-dependent switching, e.g., Guo et al., 2018, Guo et al., 2019, Guo et al., 2014 and Nie and Cao (2015). The multistability results of memristive neural networks with non-monotonic piecewise-linear activation functions are presented in Nie and Cao (2015). The existence and attractivity of multiple equilibria of memristor-based cellular neural networks with piecewise-linear activation functions are addressed in Guo et al. (2014). The multistability results of switched neural networks with piecewise-linear radial basis functions and monotonic piecewise-linear functions under state-dependent switching are presented in Guo et al., 2018, Guo et al., 2019.
As is well known, state–space partition is an important step for multistability analysis. In all existing results on multistability of switched neural networks, the activation functions considered are piecewise linear. The state–space is partitioned at the breakpoints of piecewise linear activation functions. However, state space partition is not straightforward for switched neural networks with smooth activation functions. The difficulties of multistability analysis are summarized as follows:
- (i)
State-space partition is an important step for multistability analysis. For switched neural networks with smooth activation functions, state–space partition is nontrivial.
- (ii)
Due to the discontinuity of switched neural networks, the upper and lower bound functions are also discontinuous, which leads to the difficulty on extreme point analysis.
- (iii)
Due to the discontinuity of switched neural networks, the existing analysis methods on the multistability are invalid. The existence and stability of equilibria with at least one component at a switching point need to be considered.
This paper addresses the multistability of switched neural networks with sigmoidal activation functions under state-dependent switching. By combing Brouwer fixed theorem, local linearized method, state space partition and discussion of switching threshold, sufficient conditions are derived for ascertaining the existence and stability of multiple equilibria. Compared with related existing works, the main contributions of this paper are listed as follows:
- (i)
There is no multistability results for switched neural networks with sigmoidal activation functions.
- (ii)
The number of stable equilibria of an -neuron switched neural networks is up to under given conditions.
- (iii)
Sufficient conditions are derived for ascertaining the existence and stability of the maximal number and some numbers of equilibria on and off switching points.
- (iv)
A switching rule is given for resulting in desirable number of stable equilibria.
The rest of paper is organized as follows. In Section 2, a switched neural network model with sigmoidal activation function under state-dependent switching is proposed. Then, some definitions and lemmas are presented. In Section 3, the multistability results of switched neural systems are addressed. In Section 4, four examples are given to substantiate the effectiveness of the obtained criteria. In Section 5, concluding remarks are given.
Section snippets
Preliminaries
In this section, the switched neural network model under state-dependent switching is introduced. Then, some necessary definitions and useful lemmas are also briefly outlined.
: Given the -vector , denotes the -norm of . denotes the number of elements in set . For a function , denote .
Main results
In this section, the multistability of switched neural network model (1) with activation function (2) under state-dependent switching is analyzed.
In terms of the value of switching threshold , there are three possible cases for us to discuss: Case A: , Case B: and Case C: .
To facilitate our formulation in the following discussions, a single neuron analogue (no interaction among neurons) is introduced as follows:
Numerical simulation
In this section, four illustrative examples are elaborated to substantiate the theoretical results. System (1) with two neurons has the following form: where the activation function is defined as (4) with .
First, consider Case A: .
Example 1 Consider system (32) with the parameters as follows: , ,
Concluding remarks
This paper addresses the multistability of a switched neural network model with sigmoidal activation function under state-dependent switching. By analyzing the switching threshold, partitioning state space and linearizing the system, sufficient conditions are derived for ascertaining the coexistence and stability of equilibria. Future works will aim at multistability analysis of switched neural networks with time delays and non-monotone smooth activation functions.
References (63)
- et al.
H state estimation of stochastic memristor-based neural networks with time-varying delays
Neural Networks
(2018) - et al.
Multistability and convergence in delayed neural networks
Physica D: Nonlinear Phenomena
(2007) - et al.
Identification for switched systems
IFAC-PapersOnLine
(2015) - et al.
Delay-dependent multistability in recurrent neural networks
Neural Networks
(2010) - et al.
Finite-time stabilization and boundedness of switched linear system under state-dependent switching
Journal of the Franklin Institute
(2013) - et al.
Multistability analysis of a general class of recurrent neural networks with non-monotonic activation functions and time-varying delays
Neural Networks
(2016) - et al.
Complete stability of delayed recurrent neural networks with Gaussian activation functions
Neural Networks
(2017) - et al.
Multistability of neural networks with discontinuous non-monotonic piecewise linear activation functions and time-varying delays
Neural Networks
(2015) - et al.
Simulation and stability analysis of neural network based control scheme for switched linear systems
ISA Transactions
(2012) - et al.
Pattern recognition by graph matching using the Potts MFT neural networks
Pattern Recognition
(1995)
Neural network based adaptive prescribed performance control for a class of switched nonlinear systems
Neurocomputing
Coexistence and local stability of multiple equilibria in neural networks with piecewise linear nondecreasing activation functions
Neural Networks
Exponential stability of Cohen-Grossberg neural networks
Neural Networks
Memory pattern analysis of cellular neural networks
Physics Letters. A
Associative memories based on continuous-time cellular neural networks designed using space-invariant cloning template
Neural Networks
Synchronization of complex dynamical networks with switching topology: a switched system point of view
Automatica
State-dependent switching control of switched positive fractional-order systems
ISA Transactions
State estimation of fractional-order delayed memristive neural networks
Nonlinear Dynamics
Exponential synchronization of coupled stochastic memristor-based neural networks with time-varying probabilistic delay coupling and impulsive delay
IEEE Transactions on Neural Networks and Learning Systems
Multistability of periodic delayed recurrent neural network with memristors
Neural Computing and Applications
Multiple Lyapunov functions and other analysis tools for switched and hybrid systems
IEEE Transactions on Automatic Control
Multistability in recurrent neural networks
SIAM Journal of Applied Mathematics
Multistability for delayed neural networks via sequential contracting
IEEE Transactions on Neural Networks and Learning Systems
Stability analysis and control synthesis for switched systems: A switched Lyapunov function approach
IEEE Transactions on Automatica Control
Convergence and multistability of nonsymmetric cellular neural networks with memristors
IEEE Transactions on Cybernetics
Memristor-based cellular nonlinear neural network: design, analysis, and applications
IEEE Transactions on Neural Networks and Learning Systems
Stability analysis of stochastic Cohen-Grossberg neural networks with mixed time delays
Journal of Northeastern University
Multistability of recurrent neural networks with piecewise-linear radial basis functions and state-dependent switching parameters
IEEE Transactions on Systems, Man, and Cybernetics: Systems
Multistability of switched neural networks with piecewise-linear activation functions under state-dependent switching
IEEE Transactions on Neural Networks and Learning Systems
Attractivity analysis of memristor-based cellular neural networks with time-varying delays
IEEE Transactions on Neural Networks and Learning Systems
Cited by (40)
Prediction of California bearing ratio and modified proctor parameters using deep neural networks and multiple linear regression: A case study of granular soils
2024, Case Studies in Construction Materials